Neurogenetics and brain connectomics: insights from healthy individuals, monogenic and polygenic disorders

Neu ogene ics and b ain connec omics:
insigh s om heal hy indi iduals,
monogenic and polygenic diso de s
An onio Jim´enez Ma ´ın
PhD Supe iso
Jesus M. Co es Diaz
2023
(cc)2024 ANTONIO JIMENEZ MARIN (cc by-nc 4.0)
A la amilia que engo desde que nac´ı y a la que he ido incluyendo
en es os 30 a˜nos
G acias po es a ah´ı
Resumen
La Resonancia Magn´e ica (RM) es una ´ecnica no in asi a que se
puede u iliza pa a adqui i im´agenes de al a esoluci´on de los ´o ganos
de un animal i o. Exis en una amplia a iedad de modalidades pa a
adqui i im´agenes que in oluc an di e en es aspec os de los ejidos que
o man los ´o ganos, po ejemplo, p opiedades de di usi´on del agua,
oxigenaci´on, pe usi´on, suscep ibilidad, e c. Pa a es udia el ce eb o
in i o, la esonancia magn´e ica se ha u ilizado ampliamen e en la
p ´ac ica cl´ınica y en in es igaci´on, p opo cionando in o maci´on sob e
la in eg idad del ejido ce eb al, sus mal o maciones, la o ganizaci´on
de la mic oes uc u a y la din´amica uncional.
A pa i de im´agenes de RM, el ce eb o humano se puede mod-
ela como una ed de edes, que se denomina conec oma y puede
se an o es uc u al como uncional. Es as edes, ep esen adas
ma em´a icamen e como g a os, se componen po nodos, que se co es-
ponden con las di e en es ´a eas del ce eb o que se p e ende es udia ,
y enlaces o links que de inen c´omo se conec an unos nodos con o os.
El conec oma ce eb al se puede u iliza pa a es udia poblaciones con
as o nos neu ol´ogicos o psiqui´a icos, pe o ambi´en pa a comp en-
de la o ganizaci´on a g an escala del ce eb o sano. Dado que cada
conec oma humano iene su p opia huella [1], es posible ano a sus
ca ac e ´ıs icas en unci´on de sus unciones cogni i as de al o o den,
a iables cl´ınicas, da os gen´e icos u o as ca ac e ´ıs icas eno ´ıpicas
[2]. Adem´as, las ca ac e ´ıs icas del conec oma no s´olo di ie en en e
pa icipan es sanos y pa ol´ogicos, sino que ambi´en cambian du an e
el en ejecimien o, lo que pe mi e encon a bioma cado es basados
en conec omas que pod ´ıan se ´u iles pa a el diagn´os ico y p on´os ico
de pa olog´ıas [3, 4, 5, 6, 7, 8, 9, 10]
La mul imodalidad de los da os ce eb ales adqui idos con eson-
ancia magn´e ica, jun o con o as ´ecnicas de adquisici´on, pe mi e
busca una amplia gama de bioma cado es median e en oques com-
pu acionales que ap o echan las di e en es uen es de da os. En los
´ul imos a˜nos, la publicaci´on de da os de ansc ipci´on a ni el celula
y de odo el ce eb o ha pe mi ido a los in es igado es incula las

i
ca ac e ´ıs icas del conec oma con da os gen´e icos. Es a es a egia
pe mi e iden i ica genes y ipos de c´elulas espec´ı icos implicados en
el desa ollo y p og eso de as o nos elacionados con el ce eb o [11].
T´ecnicas como los es udios de asociaci´on de odo el genoma (GWAS)
o los es udios de asociaci´on de odo el ansc ip oma (TWAS) se han
u ilizado ampliamen e pa a incula eno ipos de i ados de im´agenes
(IDP) con polimo ismos de un solo nucle´o ido (SNP), no solo en a i-
an es neu odegene a i as y neu opsiqui´a icas sino ambi´en en pobla-
ciones sanas [12, 13, 14, 15, 16, 17, 18]. Ambos en oques equie en
un g an n´ume o de pa icipan es e hip´o esis s´olidas sob e los asgos
gen´e icos que se an a es udia . Adem´as, la in o maci´on gen´e ica se
ex ae de la sang e o la sali a, po lo que no puede da in o maci´on
sob e c´omo la a ian e gen´e ica se dis ibuye espacialmen e y es ´a
a ec ada a lo la go del ejido ce eb al [11].
Un en oque al e na i o que esuel e los p oblemas an es mencion-
ados de GWAS o TWAS es el uso de a las ansc ip ´omicos ce eb ales
[19, 20, 21, 22, 23, 24], que p opo cionan la exp esi´on de miles de
genes que cub en di e en es ubicaciones del ce eb o. En e las bases
de da os disponibles, el m´as u ilizado es el Allen Human B ain A las
(AHBA), que incluye da os de seis donan es sin ninguna pa olog´ıa
conocida – po lo an o, pe mi e es udia mecanismos undamen ales
o ganiza i os – y cub e odo el ce eb o [19]. Pa a u iliza es e a las
en es udios de esonancia magn´e ica, lo que hacemos es conside a
como da os no ma i os los pe iles ansc ip ´omicos ex a´ıdos de los
ce eb os, ex ayendo odas las mues as de cada donan e en un a las
´unico. Despu´es de a ios pasos de p ep ocesamien o [25], necesa ios
pa a limpia y es anda iza los da os, es posible co elaciona la ex-
p esi´on ansc ip ´omica de los genes disponibles con los IDPs que se
desean es udia .
Hay algunas cues iones a ene en cuen a al u iliza es a base de
da os:
•Va ios genes en la base de da os ienen di e en es pa ones de
exp esi´on en e los donan es. En [26], los mismos au o es que
publica on la base de da os o iginal de inie on una medida que
iene en cuen a esa cues i´on, a sabe , la Es abilidad Di e en-
cial (DS), que indica la ep oducibilidad de la exp esi´on en los
seis donan es de cada gen. El pa ´ame o DS se puede u iliza
pa a selecciona la sonda m´as ep oducible ano ada en un gen
espec´ı ico o pa a il a genes de baja ep oducibilidad pa a los
an´alisis.
•Pa a la cons ucci´on de es e a las, s´olo se u iliza on mues as
de dos de los seis ce eb os en ambos hemis e ios. Una opci´on es
u iliza ´unicamen e mues as del hemis e io izquie do (mues eadas
en los seis ce eb os), que educe conside ablemen e los g ados
de libe ad (DOF) y, po lo an o, la signi icancia es ad´ıs ica
en los an´alisis de asociaci´on en e los IDPs y la neu ogen´e ica
[27]. O a opci´on es a˜nadi las mues as del hemis e io de echo
de los dos donan es aplicando en oques de no malizaci´on pa a
las mues as p oceden es de ambos hemis e ios [28].
•Exis e un sesgo de au oco elaci´on en e mues as ce canas –
genes m´as ce canos espacialmen e, ienen exp esi´on m´as pa e-
cida – lo que in oduce un sesgo adicional a los alo es de co -
elaci´on con los IDPs. Es e e ec o se puede educi p omediando
mues as que pe enecen a las mismas egiones del ce eb o y
eliminando los e ec os de au oco elaci´on [25, 29, 30].
Despu´es de incula el IDP de in e ´es con los da os de AHBA,
el esul ado ob enido es una lis a de genes que ienen asociaci´on sig-
ni ica i a. Es a lis a se puede analiza en p o undidad pa a ob ene
in e p e abilidad de los mecanismos neu obiol´ogicos subyacen es de
los IDPs. Po ejemplo, las es a egias basadas en hip´o esis p e ias
se cen a ´an en comp oba si la lis a de asociaci´on ob enida incluye
los genes de in e ´es elacionados con la pa olog´ıa es udiada. Al e n-
a i amen e, las es a egias basadas en da os es ´an basadas en u iliza
dis in as he amien as o en oques bioin o m´a icos, sin asumi a p io i
que cie os genes pueden ene o no mayo pa icipaci´on en la pa o-
log´ıa de es udio. De o ma o almen e di igida po da os, podemos
ob ene un anking de pa icipaci´on de genes, sob e los cuales po-
demos u iliza ´ecnicas bioin o m´a icas pa a, po ejemplo, ano a las
i
unciones biol´ogicas en las que pa icipa la lis a, u ilizando el an´alisis
de en iquecimien o (en ichmen ) de Gene On ology (GO) [31].
Los da os del AHBA se han u ilizado ampliamen e desde su pub-
licaci´on en 2012, inculando la exp esi´on ansc ip ´omica con los
IDPs elacionados con la poblaci´on sana [28, 32, 33], la en e medad
de Alzheime [34, 35], del Pa kinson [36, 37], demencia on o em-
po al [38, 39], o os as o nos neu odegene a i os [40, 41], epilepsia
[42, 43], umo es ce eb ales [44, 45], as o no del espec o au is a
[46, 27], esquizo enia [47, 48], as o no de dep esi´on mayo [49, 50]
y o os as o nos neu opsiqui´a icos [51, 52].
En es a Tesis, se ha analizado la neu ogen´e ica subyacen e a las ca-
ac e ´ıs icas del conec oma ce eb al en indi iduos sanos, en pacien es
con Dis o ia Mio ´onica Tipo I (DM1) y en indi iduos con T as o no
del Espec o Au is a (TEA).
En el cap´ı ulo 2, como cap´ı ulo me odol´ogico e in oduc o io pa a
el es o de la esis, in es igamos las elaciones en e edes ce eb ales
a m´ul iples escalas, combinando edes es uc u ales y uncionales. Es
impo an e des aca que a echa de hoy no sabemos p edeci con p e-
cisi´on la conec i idad es uc u al a pa i de la uncional, ni ice e sa.
En ealidad, es e p oblema es un desa ´ıo ac ual pa a la neu ociencia
mode na. Uno de los p incipales impedimen os es la al a de e-
posi o ios p´ublicos que in eg en edes es uc u ales y uncionales a
di e sas escalas, o esoluciones. Y es p ecisamen e aqu´ı donde hemos
que ido apo a en es e cap´ı ulo. En conc e o, pa a abo da es a
cues i´on, en es e cap´ı ulo hemos desa ollado nue os an´alisis de da os,
de c´odigo abie o que pe mi en examina la co espondencia en e las
conec i idades es uc u ales y uncionales a di e en es escalas. Hemos
desa ollado una es a egia a ni el de m´odulo, o comunidad, que
mejo a los en oques a ni el de egi´on pa a el es udio de la co es-
pondencia en e es uc u a y la unci´on ce eb al. Adem´as, ambi´en
hemos desa ollado nue os ecu sos que hacen uso de m´e icas neu o-
gen´e icas pa a el es udio de edes ce eb ales a ni el de m´odulo, que
pueden acili a nue as opo unidades pa a segui a anzando en la
in es igaci´on de neu ociencia de edes, pa icula men e en lo que e-
spec a a los as o nos ce eb ales.
ii
En el Cap´ı ulo 3, nues o obje i o ue delimi a los pe iles neu o-
gen´e icos de los pa ones de degene aci´on ce eb al en la DM1, una
en e medad monog´enica a a p oducida po la mu aci´on de un ´unico
gen (DMPK), y que en el Pa´ıs Vasco iene una p e alencia ap ox-
imadamen e es eces mayo que en o as egiones del mundo [53].
Pa a ello, c uzamos mapas ce eb ales de p´e dida de olumen (VL)
y d´e ici s neu opsicol´ogicos (ND) con los da os ansc ip ´omicos del
AHBA. Adem´as, pa a alida los esul ados ob enidos bas´andose en
da os, hemos hecho uso de es udios neu opa ol´ogicos y de ARN en
una peque˜na se ie de mues as de ce eb o con DM1.
En el Cap´ı ulo 4, hemos analizado los mecanismos molecula es
que sus en an los di e en es sub ipos de TEA, ob enidos median e
´ecnicas de clus e ing en conec omas uncionales de es os suje os.
El sub ipado en TEA es de al ´ısimo in e ´es, ya que es un as o no
del neu odesa ollo con una g an he e ogeneidad en e los genes al-
e ados, los s´ın omas de los indi iduos con TEA y sus compo ami-
en os. La es a egia seguida en es e cap´ı ulo pa a ca ac e iza es a
he e ogeneidad ha sido ag upa pe iles de conec i idad ce eb al un-
cional a g an escala (en e egiones co icales y subco icales, o mado
una ed que cub e al ce eb o comple o), en una poblaci´on de 657 in-
di iduos con au ismo. Adem´as, hemos u ilizado los da os del AHBA
pa a ca ac e iza el mecanismo molecula de ´as de cada sub ipo me-
dian e el an´alisis de en ichmen del conjun o de genes que mues an
una al a simili ud espacial con los pe iles de al e aciones de la con-
ec i idad uncional en e cada sub ipo (al e aciones medidas con e-
spec o a un g upo de pa icipan es de con ol con neu odesa ollo
no mal).
Finalmen e, y como conclusi´on de es a esis, he desa ollado di e -
en es me odolog´ıas y es a egias que se pueden aplica pa a p ocesa
da os ce eb ales de esonancia magn´e ica, con el obje i o de cons ui
conec omas es uc u ales y uncionales a mac oescala, y elaciona es-
os conec omas con da os ansc ip ´omicos espaciales de exp esi´on de
genes y con muy al a esoluci´on. La mayo ´ıa de los da os u ilizados
en es a Tesis son p´ublicos, al igual que las he amien as de las que
he hecho uso. Combinando las di e en es he amien as, he podido
xi Con en s
2.2.5 The op imal b ain pa i ion based on c oss-
modula i y maximiza ion . . . . . . . . . . . . 27
2.2.6 T ansc ip omic da a a module-le el o assess
b ain- ela ed diso de s . . . . . . . . . . . . . 27
2.3 Resul s.......................... 28
2.3.1 Op imal γ- used s uc u e- unc ion modula o -
ganiza ion.................... 30
2.3.2 Mul i-scale γ- used s uc u e- unc ion modula
o ganiza ion................... 31
2.3.3 Ana omical and unc ional desc ip ion o a gi en
b ain pa i ion . . . . . . . . . . . . . . . . . 34
2.3.4 Neu obiological ele ance o a gi en b ain pa -
i ion in b ain- ela ed diso de s by making use
o neu ogene ics da a . . . . . . . . . . . . . . 36
2.4 Discussion........................ 37
2.5 Da a and code A ailabili y . . . . . . . . . . . . . . . 41
3 T ansc ip ional signa u es o synap ic esicle genes de ine
myo onic dys ophy ype I neu odegene a ion 43
3.1 In oduc ion....................... 43
3.2 Me hods......................... 45
3.2.1 Pa icipan s, wo coho s . . . . . . . . . . . . 45
3.2.2 Demog aphic, clinical and neu opsychological
a iables..................... 47
3.2.3 MRI acquisi ion and p e-p ocessing . . . . . . 48
3.2.4 Imaging s a is ical analysis . . . . . . . . . . . 50
3.2.5 T ansc ip omics b ain maps . . . . . . . . . . 50
3.2.6 Associa ion be ween VL and ansc ip omics . 52
3.2.7 Associa ion be ween NDs and ansc ip omics 52
3.2.8 DM1 ele an genes and GO . . . . . . . . . . 53
3.2.9 Neu opa hological analyses . . . . . . . . . . . 54
3.3 Resul s.......................... 56
3.4 Discussion........................ 68

Con en s x
4 The Neu ogene ics o Func ional Connec i i y Al e -
a ions in Au ism: Insigh s F om Sub yping in 657 In-
di iduals 73
4.1 In oduc ion....................... 74
4.2 Me hods......................... 76
4.2.1 Pa icipan s . . . . . . . . . . . . . . . . . . . 76
4.2.2 Func ional Connec i i y Ma ices . . . . . . . 78
4.2.3 Da a Ha moniza ion . . . . . . . . . . . . . . 78
4.2.4 ASD Sub yping ia Consensus Clus e ing . . . 79
4.2.5 S a is ical Di e ences in B ain Mo phology and
Beha io Be ween ASD Sub ypes . . . . . . . 80
4.2.6 Func ional connec i i y al e a ions o ASD sub-
ypes wi h espec o TDC . . . . . . . . . . . 80
4.2.7 T ansc ip omics . . . . . . . . . . . . . . . . . 81
4.2.8 The use o spa ial au o eg essi e models o he
associa ion o ansc ip omics wi h sub ypes . 82
4.2.9 Gene Se En ichmen Analysis and P o ein In-
e ac ion Analysis . . . . . . . . . . . . . . . . 83
4.3 Resul s.......................... 84
4.4 Discussion........................ 95
4.5 Da a and code a ailabili y . . . . . . . . . . . . . . . 99
5 Conclusions 101
Lis o Figu es
1.1 Sagi al iew o a b ain d awing wi h ph enology-based
desc ip ion labels . . . . . . . . . . . . . . . . . . . . 2
1.2 Technologies o acqui e b ain da a a di e en spa io-
empo alscales ..................... 5
1.3 Func ional and s uc u al connec omes . . . . . . . . 6
1.4 S a e-o - he-a s- MRI p ep ocessing pipeline . . . . 11
1.5 The P inciple o DTI and Con as Gene a ion . . . . 13
1.6 FC and SC compu a ion explained . . . . . . . . . . 15
1.7 S anda d pipeline o p ep ocess ansc ip omic exp es-
sion da a om he AHBA . . . . . . . . . . . . . . . 16
1.8 Main me hods applied o s udy he spa ial associa ion
o Image De i ed Pheno ypes and gene ic in o ma ion 18
2.1 Me hodological ske ch and pipeline . . . . . . . . . . 29
2.2 Selec ion o bes ini ial pa cella ion a las and op imal
alue o γpa ame e .................. 31
2.3 G aph-node s eng h is modula ed by γ, he amoun
o in e play be ween s uc u al connec i i y (SC) and
unc ional connec i i y (FC). . . . . . . . . . . . . . . 32
2.4 Dend og am measu es and module analysis in he he
op imal b ain pa cella ion . . . . . . . . . . . . . . . 33
2.5 S a is ical dependencies be ween di e en mul i-scale
me ics.......................... 35
2.6 B ain localiza ion, Func ional desc ip ion and Ana-
omical desc ip ion o he op imal b ain pa i ion . . 36
x ii
x iii Lis o Figu es
2.7 Neu ogene ic module-le el da a o disease- ela ed an-
sc ip omic cha ac e iza ion o he op imal b ain pa i ion 38
3.1 Me hodological scheme o he associa ion be ween an-
sc ip omics and a ophy in DM1, measu ed as b ain VL 58
3.2 Da a-d i en s a egy o de e mine he associa ion be ween
he ansc ip ome and VL in DM1 . . . . . . . . . . . 60
3.3 Func ional desc ip ion o he genes wi h he highes
associa ion wi h VL . . . . . . . . . . . . . . . . . . . 61
3.4 Da a-d i en s a egy o de ine he associa ion be ween
he ansc ip ome and a en ion co-ac i a ion maps in
DM1........................... 62
3.5 Neu opa hological analyses o b ain samples e ealed
a high he e ogeneous Tau-pa hology in DM1, di e en
o heoneinAD .................... 65
3.6 RNA exp ession in hippocampal issues om DM1 pa-
ien s e ealed p o ein dys unc ion o synap ic esicle
p ocesses in DM1 . . . . . . . . . . . . . . . . . . . . 67
4.1 Gene al wo k low . . . . . . . . . . . . . . . . . . . . 86
4.2 The wo majo ASD sub ypes we e subdi ided hie -
a chically in o smalle sub ypes, bu he signi ican en-
ichmen ound only occu ed o he wo sub ypes a
he highes dend og am le el. . . . . . . . . . . . . . 87
4.3 Two majo s able ASD sub ypes, one wi h hypocon-
nec i i y and he o he wi h hype connec i i y . . . . 89
4.4 Associa ion be ween ansc ip omics and connec i i y
pa e ns o each au ism spec um diso de sub ype . 91
4.5 Exci a ion/inhibi ion imbalance en ichmen o only
one class o pa icipan s wi h au ism spec um diso de
(sub ype2) ....................... 92
4.6 Replicabili y o he esul s including samples om he
wholeb ain ....................... 94
Lis o Tables
3.1 Demog aphic, clinical and neu opsychological a iables 46
3.2 Age, gende and yea s o educa ion be ween DM1 pa-
ien sandHCs ..................... 47
3.3 Lis o ele an genes o he neu obiological aspec s o
DM1 ob ained a e e iewing p e ious s udies . . . . 55
3.4 Neu opa hological de ails o cases s udied . . . . . . . 64
4.1 Main da a cha ac e is ics o each Ins i u ion pa icip-
a ing in ou s udy . . . . . . . . . . . . . . . . . . . . 77
4.2 Beha io al Cha ac e iza ion o ASD Sub ypes . . . . 88
xix

Chap e 1
In oduc ion
1.1 The pu sui o b ain unc ions
The b ain is he mos enigma ic body o gan o any li ing animal.
In he las cen u ies, esea che s o di e en disciplines ha e ied
o explain how a bunch o neu ons could wo k oge he and how
consciousness eme ges om ha angle, de eloping a wide a ie y
o heo ies. No oo ecen , bu no oo a in ime heo ies such
as Ph enology [54], which nowadays is conside ed as pseudoscience,
we e he way o explain b ain unc ions om b ain shape. D . Gall
(1758-1826) and collabo a o s buil a ull b ain map which assigned
b ain unc ions o c anial a eas [55], guiding he clinical diagnosis by
c anial palpa ion (Figu e 1.1). These ideas se led he basis o he
B ain Localiza ion Theo y (BLT), which was de ined and de eloped
due o lesion-d i ing unc ion s udies. The majo example o he
BLT de elopmen was he disco e y o he language p oduc ion a ea
(B oca’s a ea) by D . B oca (1824-1880) [56, 57]. BLT was based on
he p inciple ha isola ed b ain egions, p oduced speci ic unc ions.
Bu , i is possible ha a single b ain egion could codi y a simple o
complex beha io ? In ha epoch, o he esea che s s a ed o ques-
ion he localiza ion heo y. D . Flou ens (1794-1867) suppo ed
ha di e en egions in e ac ed be ween hem, building unc ional
sys ems [58]. All hese wo ks we e based on he obse a ion o unc-
1
2Chap e 1. In oduc ion
ional symp oms p oduced a e b ain lesions, in humans and o he
animals.
Figu e 1.1: Sagi al iew o a b ain d awing wi h ph enology-based desc ip ion
labels. The image is ex ac ed om he book [55]. Labels meaning: a. Oli a y bodies; b.
Annula p o ube ance; c. En ance o he an e io py amids unde he annula p o ube ance;
1. Decussa ion o he an e io py amids; 48-49. Si ua ion o he o gan o philop ogeni i eness;
49-50. Si ua ion o he o gan o inhabi i eness; 50-51. Si ua ion o he o gan o sel -es eem;
51-52. Si ua ion o he o gan o i mness; 52-53. Si ua ion o he o gan o ene a ion; 53-54.
Si ua ion o he o gan o bene olence; 54-55. Si ua ion o he o gan o compa ison; 55-56.
Si ua ion o he o gan o e en uali y; I. O gan o Des uc i eness; II. O gan o ama i eness;
III. O gan o Philop ogeni i eness; IV. O gan o adhesi eness; V. O gan o inhabi i eness; VI.
O gan o comba i eness; VII. O gan o sec e i eness; VIII. O gan o acquisi i eness; IX. O gan
o cons uc i eness; X. O gan o cau iousness; XI. O gan o app oba i eness; XII. O gan o
sel -es eem; XIII. O gan o bene olence; XIV. O gan o e e ence; XV. O gan o i mness; XVI.
O gan o conscien iousness; XVII. O gan o hope; XVIII. O gan o ma ellousness; XX. O gan
o mi h ulness; XXI. O gan o imi a ion; XXVI. O gan o colo ing; XXVII. O gan o locali y;
XXVIII. O gan o o de ; XXIX. O gan o calcula ion; XXX. O gan o e en uali y; XXXII.
O gan o une.
Du ing 19 h and 20 h cen u y, BLT heo y was ex ensi ely used in
clinical p ac ice, and e y epu ed clinicians and esea che s, such as
D . Ka l We nicke (1848-1905), D . John Hughlings Jackson (1835-
The pu sui o b ain unc ions 3
1911), o D , Wilde Pen ield (1891–1976), we e mapping he b ain
a eas wi h hei unc ions. Pa ien s as HM (1926-2008), who had a
ull memo y consolida ion impai men due o su gical ejec ion o he
hippocampus, helped he es ablishmen o BLT heo y.
Despi e conside able ad ances, he b ain emains a mys e y, bu
wi h he ac ual echnology, esea che s ad anced mo e quickly in he
solu ion o his unknown and a e mo e ce ain ha b ain unc ions a e
ela ed o ne wo ks, bu no o speci ic b ain a eas. So, o example,
B oca’s a ea is a pa (maybe he mos impo an ) o he language
ne wo k.
Recen ly, a new heo y de eloped by D . Michael D. Fox and
collabo a o s, named Lesion Ne wo k Mapping (LNM), has come up
s udying also b ain lesions and he symp oms which p oduced [59].
LNM is based on he s udy o which b ain ne wo ks a e connec ed o
he lesioned a ea in a la ge no ma i e da ase , hus he pa ien wi h
ha lesion would ha e po en ially a ec ed o disconnec ed hose ne -
wo ks. Following he same p ocess in a g oup o pa ien s wi h he
same neu ologic o psychia ic symp oms, bu wi h di e en lesioned
b ain egions, would de ine he common ne wo k ha has been dis-
connec ed in all he pa ien s, which would be he one p oducing he
symp om. LNM has been p o ed in nume ous quan i y o neu olo-
gic, mo o and non-mo o , beha iou al, and psychia ic symp oms
(see [60] and e e ences he ein). Fu he mo e, i has been es ed o
i s po en ial use o deep b ain s imula ion [61]. Following his, e-
sea che s such as D . Mau izio Co be a, and also ou g oup, ha e
also applied LNM o mapping cogni i e unc ions o b ain ne wo ks
h ough b ain lesions [62, 63, 64]. LNM is a p omising echnique ha
inc eases he ce ain y o cogni i e unc ions a e he esul o ne wo k
ope a ions inside he b ain.
Fo ha eason, in his Thesis, he s a ing poin is he unde -
s anding o he b ain as a complex ne wo k.
10 Chap e 1. In oduc ion
•Dis o ion co ec ion: he sequence used o acqui e MRI
images is g adien -echo echoplana imaging (EPI) and p oduces
a e ac s whe e ai and issue mee , e.g. he sinuses (p oducing
a e ac s in he o bi o on al a ea) and he ea s (p oducing a e-
ac s in he empo al poles) [68]. To co ec hese e ec s, i is
necessa y o acqui e an addi ional sequence called ield map
which cha ac e izes he MRI scanne ield.
•Remo al o unwan ed signals and ends: despi e he
BOLD signal being ela ed o neu al ac i i y p oduced in he
GM, he e a e also signals in WM and CSF. These signals a e
usually conside ed as nuisances o emo e [71] oge he wi h
linea and quad a ic ends, using a gene alized linea model
(GLM).
•Tempo al il e ing: he s- MRI BOLD signals ha e hei
physiological cha ac e is ics a equencies be ween 0.01-0.08Hz
[72], so a bandpass il e will be equi ed in ha ange.
•Regis a ion o a common empla e: his s ep is no man-
da o y, bu i is needed o g oup analyses, egis e ing he indi-
idual images o a common empla e and space. The common
empla e could be he one buil wi h he indi idual ana omical
images o a no ma i e ana omical empla e, e.g. MNI152.
•Spa ial smoo hing: a spa ial smoo hing, usually a 3D Gaus-
sian il e , is used o educe he noise o his sequence and o
inc ease he s a is ical powe in g oup compa ison s udies, as
smoo h educes indi idual a iabili ies. The il e mus be smal-
le han he ac i a ion signal o be de ec ed, and usually is 2
imes he size o he oxel [68].
All hese s eps mus be pe o med sequen ially and o de ed. In
igu e 1.4 a s a e-o - he-a pipeline shows an example o he di e en
s eps and he o de ha should be ollowed:
The e exis s comple e s- MRI p ep ocessing pipelines commonly
used by he neu oimaging communi y such as m ip ep [73] h ps://

MRI da a p ep ocessing 11
Figu e 1.4: S a e-o - he-a s- MRI p ep ocessing pipeline.
m ip ep.o g/en/s able/,CONN [74] h ps://web.conn- oolbox.o g/, o C-PAC
[75] h ps://gi hub.com/FCP-INDI/C-PAC. In his Thesis, I de eloped and
eely sha ed ou own MRI p ep ocessing pipeline which adjus ed
be e o ou needs and lab acili ies as he men ioned ones. I can
be ound a : h ps://gi hub.com/compneu obilbao/compneu o- m ip oc
1.5.3 DTI p ep ocessing
In his sequence, he p ep ocessing no only includes he a e ac e-
mo al bu also he i ing o he di usion enso and he ib e acking.
The main a e ac s a e p oduced by he dis o ions o he magne ic
ield, eddy cu en s, and head mo emen .
•Eddy cu en co ec ion: he s ong magne ic ield g adien s
applied o acqui e he enso imaging may lead o eddy cu en s
which p oduce s e ches, shea , alse ib e acking, enhaced
backg ound, image in ensi y loss, and image blu ing. The main
ool o co ec hese dis o ions and also o head mo ion is he
eddy ool included in FSL.
12 Chap e 1. In oduc ion
•Dis o ion co ec ion: simila o he MRI sequence (bo h
MRI and DTI a e called Echo-Plana imaging (EPI) sequences)
is a ec ed by simila ypes o a e ac s. To co ec hem, one
can use ield map images and also i is necessa y o acqui e wo
e e ence images wi h con a y o ien a ions. FSL so wa e also
includes he ool opup o co ec ing hem.
•Di usion enso i ing: o each oxel he mo emen o he
wa e molecules can be desc ibed using a enso (Figu e 1.5
panel B). A local i ing o his enso can be compu ed on he
co ec ed image. Di e en measu es such as F ac ional Aniso-
opy (FA, Figu e 1.5 panel D), mean di usi i y (MD), Radial
Di usi i y (RD), and Axial Di usi i y (AD), can be calcula ed
using he eigen ec o s and eigen alues ob ained.
•Fib e acking: knowing he di ec ion and o ien a ion o he
enso in each oxel (Figu e 1.5, panel E) we can ollow a pa h-
way ha connec s wo b ain egions in a p ocess called ac o-
g aphy. The e exis wo kinds o ac og aphy algo i hms: de-
e minis ic and p obabilis ic. The de e minis ic ac og aphy
p ocess [77] s a s wi h ixing a seed in a selec ed loca ion and
acking he indi idual ib es by connec ing he oxels wi h he
adjacen ones owa d he ib e di ec ion. Tha ib e di ec ion
is de e mined by he leading local eigen ec o o he di usion
enso (Figu e 1.5 panel C). This is an i e a i e p ocess un-
il a e mina ion c i e ion is eached, such as minimum FA,
ib e leng h, ib e cu a u e, o e mina ion poin a i ing. In
p obabilis ic ac og aphy he e is also a andom p ocess in he
p opaga ion di ec ion. The way he ib e is p opaga ed can
a y depending on he algo i hm used, he mos popula a e:
p ob ackx [78], iFOD1 [79], and iFOD2 [80].
Common so wa e o DTI p ep ocessing a e FSL,MR ix3 (h ps:
//m ix. ead hedocs.io/en/de /index.h ml), and DIPY (h ps://dipy.o g/). The e
also exis s a s anda d p ep ocessing pipeline DWIp ep (h ps://gi hub.
com/GalKeple /dwip ep) which uses FSL and MR ix3 o doing he s eps.
B ain ne wo ks 13
Figu e 1.5: The P inciple o DTI and Con as Gene a ion. This igu e is ex ac ed
om [76]. ”F om di usion measu emen s along mul iple axes (A), he shape and he o ien a ion
o a “di usion ellipsoid” is es ima ed (B). This ellipsoid ep esen s wha an ink s ain would be
i ink we e d opped wi hin he pixel. An aniso opy map (D) can be c ea ed om he shape,
in which da k egions a e iso opic (sphe ical) and b igh egions a e aniso opic (elonga ed).
F om he es ima ed ellipsoid (B), he o ien a ion o he longes axis can be ound (C), which is
assumed o ep esen he local ib e o ien a ion. This o ien a ion in o ma ion is con e ed o a
colo (F) a each pixel. By combining he in ensi y o he aniso opy map (D) and colo (F), a
colo -coded o ien a ion map is c ea ed (E).”
In his Thesis, I de eloped and eely sha ed ou own DTI p ep o-
cessing pipeline which adjus ed be e o ou needs and lab acili ies.
I can be ound a : h ps://gi hub.com/compneu obilbao/compneu o-dwip oc
1.6 B ain ne wo ks
The human connec ome is modeled as a ne wo k o ne wo ks, which
a e complex sys ems composed o simila pa s, named nodes, in e -
connec ed be ween hem by edges, allowing he in o ma ion low in
he whole sys em. In igu e 1.3, panels B and C, nodes a e he colo ed
14 Chap e 1. In oduc ion
do s ha ep esen he b ain egions o panel A, and edges a e he
links in blue-g een colo s connec ing he nodes. The ma hema ical
language ha desc ibes and quan i ies ne wo ks is called g aph heo y.
Selec ing he nodes o he b ain g aph is a c i ical s ep o he
analyses one wan s o pe o m. One al e na i e could be o selec all
he oxels o small pa ches con aining se e al oxels [2], bu i could
be compu a ionally expensi e. Al e na i ely, i is common he use o
b ain pa cella ions o a lases ha can be ei he unc ional [81, 82],
s uc u al [83, 84], o bo h s uc u al and unc ional [85, 86, 87].
Finally, i is also possible o apply a da a-d i en s a egy o clus e
b ain mic o- egions based on he homogenei y o he da a o ce ain
cha ac e is ics o he ne wo ks [2].
Once he b ain-nodes a e selec ed, he nex s ep is o de ine he
connec i i y be ween hem. Two main classes o connec i i y exis :
•Func ional connec i i y (FC): desc ibes he ac i a ion p o-
iles and dependencies among dis inc , and may be dis an ,
neu al popula ions o b ain egions [88]. The s onge he s a -
is ical dependency be ween bo h a eas, he highe will be he
FC alue o he edge. Tha dependency could be measu ed by
di e en me hods, such as Pea son’s Co ela ion, pa ial co el-
a ion, mu ual in o ma ion, e c (Figu e 1.6, panel A).
•S uc u al connec i i y (SC): desc ibes he ana omical con-
nec ions be ween b ain egions h ough axon bundles o WM
ac s [2]. Usually he measu e u ilized is he numbe o ib es
connec ing wo egions (Figu e 1.6, panel B).
Ano he way o es ima e connec i i y edges is he E ec i e con-
nec i i y (EC) ha desc ibes causal in e ac ions unde lying he in-
o ma ion low [2], bu his class has no been used in his Thesis.
Neu ogene ics and connec omics 15
Figu e 1.6: FC and SC compu a ion explained.A: wo ime-se ies (he e Thalamus and
P ecen al) a e co ela ed be ween hem using Pea son’s co ela ion, he esul is s o ed in he
FC ma ix. The es o he FC alues a e om he co ela ions be ween all he egions (in
pai s) included in Desikan’s pa cella ion. B: The numbe o ib es (# ) be ween Thalamus
and P ecen al is s o ed in he SC ma ix. As in FC, he es o he alues a e om he #
be ween all he egions included in Desikan’s pa cella ion. No ice ha , in he SC hea map #
is in logscale
1.7 Neu ogene ics and connec omics
In he las yea s, he elease o b ain-wide and single-cell ansc ip ion
da ase s has enabled esea che s o link connec ome cha ac e is ics
wi h gene ic da a, o a e iew see [11]. This s a egy allows o
iden i ying speci ic genes and cell ypes implica ed in he de elopmen

16 Chap e 1. In oduc ion
and p og ess o b ain- ela ed diso de s [11].
Techniques such as Genome Wide Associa ion S udies (GWAS) o
T ansc ip ome Wide Associa ion S udies (TWAS) ha e been b oadly
used o link Image De i ed Pheno ypes (IDPs) wi h Single Nucleo ide
Polymo phisms (SNPs), no only in neu odegene a i e and neu opsy-
chia ic a ian s bu also in heal hy popula ions [12, 13, 14, 15, 16,
17, 18]. Bo h app oaches equi e a la ge numbe o pa icipan s and
s ong hypo heses on he gene ic ai s o be s udied. Fu he mo e,
he gene ic in o ma ion is ex ac ed om blood o sali a, so i can-
no gi e in o ma ion abou how he gene ic a ian impac s he b ain
issue [11].
An al e na i e app oach ha sol es he a o emen ioned issues o
GWAS o TWAS is he use o b ain- ansc ip omic a lases [19, 20,
21, 22, 23, 24], which p o ide he exp ession o housands o genes
co e ing di e en b ain loca ions. Among he a ailable da ase s, he
mos used is he Allen Human B ain A las (AHBA), which includes
da a om six dono s wi hou any known pa hology and co e s he
ull b ain [19]. Fo using his a las in MRI s udies, we conside as
no ma i e da a he ansc ip omic p o iles ex ac ed om he b ains,
pulling all he samples om each dono in a unique a las. A e
se e al p ep ocessing s eps [25], needed o clean and s anda dize he
da a, i is possible o co ela e he ansc ip omic exp ession o he
a ailable genes o he IDPs ha one wan s o s udy.
Figu e 1.7: S anda d pipeline o p ep ocess ansc ip omic exp ession da a om
he AHBA. The igu e is ex ac ed om [25]
Neu ogene ics and connec omics 17
In his Thesis, I used he AHBA o he neu ogene ic analyses in
chap e s 2, 3, 4.
The e a e some issues o ake in o accoun when using his da ase :
•Se e al genes in he da abase ha e di e en exp ession pa e ns
ac oss he dono s. In [26], he same au ho s o he o iginal da a-
se de ined a measu e accoun ing o ha issue. The Di e en-
ial S abili y (DS) ells he exp ession ep oducibili y in he six
dono s o each gene. DS can be used o selec ing he mos
ep oducible p obe anno a ed o a speci ic gene o o il e ing
low- ep oducible genes o he analyses.
•Only wo o he six b ains we e sampled in bo h hemisphe es.
One op ion is o only use samples om he le hemisphe e
(sampled in he six b ains), bu i will educe he Deg ees O
F eedom (DOFs) o he associa ion analyses be ween IDPs and
neu ogene ics, hus educing he signi icance [27]. Howe e , i is
possible o add he igh hemisphe e samples o he wo dono s
applying sample no maliza ion app oaches [28].
•The e exis s an au oco ela ion bias be ween nea by samples,
which in la es he co ela ion alues wi h he IDPs. This e ec
can be educed by a e aging samples belonging o he same
b ain egions and emo ing he au oco ela ion e ec s [25, 29,
30].
A e linking he IDP o in e es wi h he AHBA da a, he ou -
come will be he lis o signi ican associa ed genes. Tha lis can be
u he analyzed o gain in e p e abili y o IDP unde lying neu obi-
ological mechanisms. Fo example, hypo hesis-d i en s a egies will
ocus on checking i genes o in e es ela ed o he pa hology s ud-
ied a e included in he lis . By con as , da a-d i en s a egies can
be pe o med using dis inc bio-in o ma ics ools o app oaches. Fo
example, i is possible o anno a e he biological unc ions ha he
lis is pa icipa ing in, using he Gene On ology (GO) en ichmen
analysis [31]. Figu e 1.8 shows a summa y o how o link AHBA da a
o a IDP.
18 Chap e 1. In oduc ion
Figu e 1.8: Main me hods applied o s udy he spa ial associa ion o Image De i ed
Pheno ypes and gene ic in o ma ion. The igu e is ex ac ed om [11]. ”T ansc ip ome
da a is ob ained om pos mo em b ain issue om di e en b ain egions ac oss he co ex (A,
AHBA example). The ansc ip ome ma ix (b ain samples x genes) con ains he exp ession o
all he measu ed genes o each sample o he b ain. Compu ing he simila i y o he spa ial
exp ession o each pai o genes in he b ain he gene-coexp ession ma ix is ob ained, which
iden i ies g oups o genes wi h simila spa ial pa e ns. Single cell ansc ip ome da a allow o
classi y each gene in o a cell ype (B). This in o ma ion allows us o compu e he dis ibu ion
o he spa ial simila i y o he IDP wi h he genes pe aining o each cell ype. The spa ial
simila i y o he whole ansc ip ome can be compu ed wi h he IDPs o ob ain a dis ibu ion o
simila i y alues (C has been adap ed om [28], and C-F shows he signi ican old en ichmen
on Posi i e Regula ion o Synapse Assembly as a case example). The op anked genes (mos
simila o he IDPs) can be used in a gene-se en ichmen analysis o ind signi ican ly associa ed
unc ional anno a ions wi h he genes (C). The unc ional anno a ions o genes o in e es can
be u he explo ed o ob ain he co ela ion alue wi h IDP, p ojec hei alues on o he b ain
su ace and compu e he empo al ajec o y along he li espan (D-E). This op anked lis can
also be used o ind ela ionships wi h o he genes and ind he impo ance o ou candida e
genes in he ob ained in e ac omic gene ne wo k (F). The posi ion o each node in he ne wo k is
plo ed as a unc ion o he hubness o each gene (o deg ee cen ali y, he numbe o connec ions
o o he genes o he ne wo k) in he ba s igu e in F.”
Thesis o e iew 19
AHBA da a has been b oadly used since i s publica ion in 2012,
linking he ansc ip omic exp ession wi h IDPs ela ed o heal hy
popula ion [28, 32, 33], Alzheime ’s disease [34, 35], Pa kinson’s dis-
ease [36, 37], F on o- empo al Demen ia [38, 39], o he neu odegen-
e a i e diso de s [40, 41], Epilepsy [42, 43], b ain umou s [44, 45],
Au ism Spec um Diso de [46, 27], Schizoph enia [47, 48], Majo
Dep ession Diso de [49, 50], and o he neu opsychia ic diso de s
[51, 52].
1.8 Thesis o e iew
In his Thesis, I ha e analysed he neu ogene ics unde lying b ain ne -
wo k cha ac e is ics in heal hy indi iduals, in pa ien s wi h Myo onic
Dys ophy Type I (DM1), and in pa ien s wi h Au ism Spec um Dis-
o de (ASD).
Chap e 2 con ains a s uc u al- unc ional mul i-scale app oach
o making b ain pa cella ions a di e en ne wo k esolu ions. The
app oach also includes a amewo k o ob ain he gene exp ession o
each ROI o he pa cella ion using he AHBA da ase . Fu he mo e,
o he op imal b ain pa cella ion, i is de ailed he gene ic ele ance
o each module in a se o se e al b ain- ela ed diso de s.
In Chap e 3, he objec i e was o delinea e he neu ogene ic p o-
iles o b ain degene a ion pa e ns in DM1, a a e monogenic dis-
o de p oduced by a mu a ion o a single gene (DMPK ), and ha in
he Basque Coun y has a p e alence abou h ee imes highe han
o he egions in he wo ld [53]. To do ha , we in e sec b ain maps
o olume loss (VL) and neu opsychological de ici s (NDs) wi h he
ansc ip omic da a om he AHBA. Fu he mo e, o alida e he
esul s ob ained in a da a-d i en way, neu opa hological and RNA
analyses we e pe o med in a small se ies o DM1 b ain samples.
In Chap e 4, we analysed he molecula mechanisms behind wo
sub ypes o ASD, which is a neu ode elopmen al diso de wi h a la ge
he e ogenei y among he genes al e ed, pa ien ’s symp oms, and be-
ha iou s. We applied a sub yping app oach based on consensus clus-
e ing o unc ional b ain connec i i y pa e ns o a popula ion o 657
26 Chap e 2. Linking s uc u e, unc ion and neu o-gene ics
ac oss all equi alen links in he indi idual ma ices. Nex , and o
ma ch he spa se na u e o SCp, a h eshold was applied o FCp,
esul ing in compa able link densi y. Subsequen ly, we bina ized he
wo ma ices and used hem in o a single ma ix as ollows:
γSFC ≡γ×FCp+ (1 −γ)×SCp,
whe e γis a eal numbe be ween 0 and 1. This s a egy allows us
o con inuously pa ame e ize he le el o in e play be ween SCpand
FCp, enabling he eco e y o each connec i i y class when γ= 0 o
γ= 1, while ob aining hyb id connec i i ies o in e media e alues.
2.2.4 Mul i-scale hie a chical ep esen a ion a
module-le el
Fo each o he 9 dis inc iPAs, a hie a chical agglome a i e clus e ing
(HAC) was employed o ex ac di e en nes ed modules, allowing o
build di e en ne wo k esolu ions a di e en scales (each one de-
e mined by a cu ing ee numbe o modules M). Mo e speci ically,
HAC was pe o med using a weigh ed me hod on he connec i i y
pa e ns ob ained om he ma ix γSFC, he e de ined as he co el-
a ion dis ance be ween pai s o mic o- egions in he γSFC ma ix.
Wi hin he con ex o ee o dend og am pa i ions, di e en
mul i-scale me ics can be es ablished. These me ics can be de ined
ei he a he module-le el o a a mo e g anula le el (wi h he highes
spa ial esolu ion de e mined by he mic o- egion le el). In his pa -
icula s udy, we ha e de ined he module size (MS), ep esen ing he
coun o mic o- egions encompassed wi hin a module. Addi ionally,
we ha e in oduced he mul i-scale index (MSI), which quan i ies he
numbe o le els wi hin he ee s uc u e whe e a module emains
in ac wi hou u he di ision. In ela ion o mic o- egions, we ha e
selec ed he heigh (H) me ic, which deno es he speci ic le el a
which a gi en mic o- egion sepa a es om i s pa en module.

Ma e ials and Me hods 27
2.2.5 The op imal b ain pa i ion based on c oss-
modula i y maximiza ion
A e ob aining a hie a chical pa i ioning o he connec i i y ma ix,
clus e ing mic o- egions in o dis inc modules M, he de e mina ion
o he op imal le el o cu ing he ee depends on he speci ic me ic
being op imized. In his s udy, ollowing he me hodology in oduced
in [86], he me ic chosen o maximiza ion is he c oss-modula i y χ,
de ined as:
χ≡(QF×QS×TF S)1
3
This me ic simul aneously accoun s o h ee di e en quali -
ies: he modula i y o he unc ional ma ix (QF), he modula i y
o he s uc u al ma ix (QS), and hei simila i y (TF S ). The la -
e is de ined as he DICE simila i y be ween a unc ional module
and a s uc u al module, and a e aging ac oss all modules in he
gi en pa i ion. By a ying M along he ee, we can calcula e χ o
each con igu a ion M and selec he op imal pa i ion M∗whe e χis
maximized. Addi ionally, unlike he app oach in [86], we he e ha e
in oduced a second pa ame e γ∗in he maximiza ion o χ, which
con ols he amoun o s uc u e- unc ion in e play.
2.2.6 T ansc ip omic da a a module-le el o as-
sess b ain- ela ed diso de s
In addi ion o mul i-scale b ain pa i ions and s uc u e- unc ion ana-
lysis, we p ocessed he ansc ip omic open da a om he Allen Hu-
man B ain A las (AHBA) [19] by using he abagen ool de eloped in
[129]. In pa icula , abagen allows us he gene a ion o b ain an-
sc ip ome module-le el alues o a speci ic pa cella ion, ha in ou
case, we chose he op imal b ain pa i ion. Mo eo e , we examined
he ansc ip omic exp ession a module-le el in ela ion o genes as-
socia ed wi h 32 b ain diso de s in oduced in [133]. To do his, we
calcula ed he mean ansc ip omic exp ession o all genes ela ed
o each diso de wi hin each module. Subsequen ly, we pe o med a
28 Chap e 2. Linking s uc u e, unc ion and neu o-gene ics
z-sco e analysis o each diso de o iden i y modules wi h low an-
sc ip omic exp ession (z <-2) and modules wi h high ansc ip omic
exp ession (z >2). The a e age exp ession o he di e en genes
we e also g ouped in o 7 di e en disease ca eogo ies: Psychia ic
diso de s, Subs ance abuse, Mo emen diso de s, Neu odegene a i e
diseases, Tumo condi ions, o De elopmen al diso de s.
2.3 Resul s
S uc u al and unc ional connec i i y ma ices a a ious esolu ions
we e buil making use o b ain images om he open da ase “Max
Planck Ins i u Leipzig Mind-B ain-Body Da ase ” – LEMON [131],
well-known o ha ing high-quali y mul imodal acquisi ions and p e-
p ocessed MRI sequences. Ra he han using di e en subjec s ac oss
li espan, we speci ically selec ed 136 young pa icipan s, aged be ween
20 and 30 yea s old, who had bo h he p ep ocessed s- MRI se-
quence and DWI sequence a ailable. We nex p ocessed he aw im-
ages ollowing s anda d neu oimaging pipelines o ob ain s uc u al
connec i i y (SC) and unc ional connec i i y (FC) ma ices (Fig-
u e 2.1A). Subsequen ly, we buil popula ion connec i i y ma ices
by choosing, o each link in he ma ix, he median alue ac oss all
he equi alen links in he indi idual connec i i y ma ices. Finally,
we modeled he amoun o in e play be ween s uc u e and unc ion
connec i i ies using a single usion pa ame e , γ. When γequals 0 o
1, we ha e pu ely s uc u al o unc ional connec i i y, espec i ely.
In in e media e si ua ions, he amoun o o e lapping connec i i y
is modula ed by γ. Fu he mo e, and simila o he me hodology
desc ibed in [86], o each alue o he usion pa ame e γ, we ap-
plied hie a chical agglome a i e clus e ing o he esul ing γ- used
s uc u e- unc ion ma ix, γSFC (Figu e 2.1B).
Resul s 29
Figu e 2.1: Me hodological ske ch and pipeline.A: Pa icipan s’ aw MRI da a om
N=136 heal hy olun ee s o he MPI-LEMON da ase we e used. F om p ep ocessed s- MRI
images, we gene a ed an ini ial b ain pa cella ion using g ay-ma e masks and he open epos-
i o y pyClus e ROI. Subsequen ly, we employed ou own publicly a ailable code on Gi Hub o
p ocess he DWI images and ex ac he SC ma ices. All pos -p ocessed da a a he pa icipan
le el, including he ime se ies BOLD signal co esponding o he di e en iPAs, and he SC and
FC ma ices a e a ailable a h ps://zenodo.o g/ eco d/8158914.B: We calcula ed popu-
la ion SCpand FCpma ices, and subsequen ly gene a ed γ- used ma ices as desc ibed in he
me hods sec ion. The γSFC ma ices we e used o cons uc dend og am ees o di e en γ
pa ame e s. Addi ionally, we employed abagen o gene a e ansc ip omic exp ession ma ices
o a ious ee-based pa cella ions. The ansc ip omic ma ices and dend og am ees a e
also accessible on Figsha e. The en i e pipeline and p ojec codes can be ound on Gi Hub a
h ps://gi hub.com/compneu obilbao/bha2.
30 Chap e 2. Linking s uc u e, unc ion and neu o-gene ics
2.3.1 Op imal γ- used s uc u e- unc ion modu-
la o ganiza ion
The di e en ees o he γSFC ma ices desc ibe dis inc scales, en-
abling he cons uc ion o ne wo ks a di e en esolu ions. The de-
e mina ion o he op imal pa i ion by cu ing he ee a M∗mod-
ules depends on he speci ic me ic we seek o maximize. In o de o
showcase he e sa ili y o ou da a wi h a ious me ics, we speci ic-
ally concen a e on a me ic we ha e p e iously de ined and named
c oss-modula i y [86]. Rep esen ed as χ, i is de ined as he p oduc
o h ee quan i ies: he modula i y o he unc ional ma ix, he mod-
ula i y o he s uc u al ma ix, and he a e age simila i y be ween
s uc u al and unc ional modules.
The e o e, we aimed o maximize χ o selec he op imal alue o γ
o a gi en ini ial pa cella ion a las (iPA). Figu e 2.2 shows box-plo s
o di e en alues o he c oss-modula i y χac oss di e en dend o-
g am le els and o di e en iPAs wi h di e en size, ie., a ying he
numbe o mic o- egions anging om 183 o 2165. Fo u he ana-
lyses, we selec ed he iPA wi h 2165 mic o- egions since i exhibi ed a
highe mean alue oge he wi h a lowe a iabili y alue o χwi hin
a wide ange o scales, pa icula ly wi hin he i s 120 le els whe e
he maximum χis loca ed. Beyond his poin , he e was a dec eas-
ing end. The op imal b ain pa i ion co esponds o he iPA ha
maximizes χby educing he dend og am o M∗modules and se ing
he usion pa ame e o γ∗. In ou case, he op imal alue γ∗was
de e mined o be 0.7, and his occu ed a he le el o 28 modules.
Howe e , wo modules we e conside ed in alid as hey consis ed o
only one and wo mic o- egions, espec i ely, making i imp ac ical
o analyze he communi y s uc u e wi hin hem. Hence o h, any
subsequen analysis e e ing o he op imal b ain pa i ion will be
based on he iPA ha maximizes χby educing he dend og am o
M∗= 26 modules and se ing he usion pa ame e o γ∗= 0.7.
Resul s 31
Figu e 2.2: Selec ion o bes ini ial pa cella ion a las and op imal alue o γpa a-
me e .Le : Box-plo s o he c oss-modula i y χac oss di e en dend og am le els and o
di e en size o he ini ial pa cella ion a las (iPA). The di e en alues o χ, on which he box-
plo s a e calcula ed, come om he di e en le els in he dend og am (he e we ha e a ied om
2 o 120). As he numbe o ROIs o he ini ial pa cella ion inc eases, he c oss-modula i y χ
also inc eases. Righ : The op imal alue o γ∗= 0.7 is also illus a ed o he same ange o
dend og am le els, ie. om 2 o 120.
2.3.2 Mul i-scale γ- used s uc u e- unc ion mod-
ula o ganiza ion
As an illus a i e example, and o gain a deepe unde s anding o how
he mul i-scale o ganiza ion is in luenced by γ, we in es iga ed he
ne wo k s eng h o he γSFC ma ix in wo ne wo k ep esen a ions.
A he ines spa ial esolu ion, co esponding o he lowes le el in
he hie a chical ee consis ing o 2165 mic o- egions ( ep esen ed as
b ain maps in Figu e 2.3A), and o each o he mac o- egions, in-
cluding he on al lobe, pa ie al lobe, occipi al lobe, empo al lobe,
insula, and a collec ion o subco ical a eas (Figu e 2.3B). F om bo h
analyses, i is e iden ha inc easing γ, ansi ioning om s uc u e
(γ= 0) o unc ion (γ= 1), esul s in highe s eng hs shi ing om
subco ical a eas o co ical egions. This pa e n is obse ed con-
sis en ly ac oss all mac o- egions, whe e he s eng h p og essi ely
inc eases wi h γ, peaking a ound γ≈0.7.

32 Chap e 2. Linking s uc u e, unc ion and neu o-gene ics
Figu e 2.3: G aph-node s eng h is modula ed by γ, he amoun o in e play
be ween s uc u al connec i i y (SC) and unc ional connec i i y (FC). A: B ain
maps o g aph-node (no malized) s eng h. I is obse ed ha o γ= 0 (pu e SC), high
s eng hs a e p edominan ly localized in subco ical egions. Howe e , as γ a ies, he e is a
dec ease in s eng h wi hin hese egions and an inc ease in co ical egions. B: Box plo s o
he dis ibu ion o g aph-node s eng h ac oss di e en mac o- egions: F on al lobe, Pa ie al
lobe, Tempo al lobe, Occipi al lobe, a collec ion o Subco ical a eas, and Insula. No ably,
o each mac o- egion, he s eng h shows an inc easing end om γ= 0, eaching a peak a
γ≈0.7, ollowed by a subsequen dec ease.
Beyond he wo selec ed le els o esolu ion o calcula ing s eng h,
a hie a chical ee o ganiza ion enables ob aining di e en me ics a
di e en module-le els. Fo ins ance, we i s colo ed he M∗= 26
dis inc modules o he op imal b ain pa i ion o ease o isualiz-
a ion (Figu e 2.4, op panel). Some ee me ics we compu ed we e
he module size (MS), heigh (H), and mul i-scale index (MSI). While
H is de ined o each mic o- egion, MS and MSI a e de ined a he
module-le el.
Resul s 33
Figu e 2.4: Dend og am measu es and module analysis in he he op imal b ain
pa cella ion. Each colo in he dend og am co esponds o a dis inc module ( ha o he
op imal b ain pa i ions co espond o 26 di e en modules). The y-axis o he dend og am
illus a es di e en le els (L) and p o ides isual ep esen a ion o a gi en module and hei
in e io (Lin ) and supe io (Lsup) le els. Di e en mul i-scale me ics can be de ined, eg.,
highe alues o he Mul i-scale Index (MSI) co espond o enhanced p ese a ion o he module
h oughou mul iple le els in he dend og am, indica ing heigh ened s abili y ac oss he ee
s uc u e. The module size is also ep esen ed by he wid h o each module in he dend og am.
The lowe inse shows a ma ked mic o- egion Heigh indica ing o which ex en a gi en mic o-
egion is de ached om he o e all ee s uc u e. Lowe alues o Heigh ep esen s onge
connec i i y o he mic o- egion wi h he o he mic o- egions belonging o he same module.
The igu e also shows b ain maps depic ing he h ee a o emen ioned measu es o each o he
modules.
We also de ined he module heigh (MH), by a e aging o e all
indi idual H alues wi hin he gi en module (Figu e 2.4, ull dend o-
34 Chap e 2. Linking s uc u e, unc ion and neu o-gene ics
g am and dend og am-inse ). We ep esen ed hese me ics on b ain
plo s, whe e we obse ed dis inc pa e ns o each module (Figu e
2.4, b ain plo s).
We conduc ed u he analyses o examine he co ela ions among
hese me ics and he in a-module s eng h, as a p oxy o module
seg ega ion (Figu e 2.5). F om he di e en ee measu es, bo h MS
and MH showed a high co ela ion ( = 0.95, p <0.001) wi h module
seg ega ion, as well as be ween hemsel es ( = -0.84, p <0.001).
In e es ingly, he i s module M1 was an ou lie in he me ic o MSI.
Loca ed a he bo de be ween he P ecuneus, Is hmus Cingula e, and
Pos e io Cingula e co ices (Figu e 2.6), M1 showed a ema kable
esilience in emaining in ac wi hou spli ing ac oss mul iple le els
in he ee, possibly indica ing a mul i-scale unc ional ole.
2.3.3 Ana omical and unc ional desc ip ion o a
gi en b ain pa i ion
Ou me hodology o de ining a gi en pa i ion de ined by speci ic
alues o M and γis gene ally da a-d i en. A some poin , i may be
o in e es o cha ac e ize he ana omical and unc ional p ope ies
o a speci ic pa i ion. As an illus a i e example, we made use o he
op imal pa i ion wi h M∗= 26 modules and γ∗= 0.7 and de e m-
ined he spa ial loca ions o hese 26 modules wi hin he b ain (Figu e
2.6, b ain-glasses). We nex ob ained hei unc ional and ana om-
ical cha ac e iza ion by measu ing espec i ely he amoun o o e lap
wi hin known Res ing S a e Ne wo ks (RSNs) [81] and egions om
he Desikan-Killiany a las [83], he la e consis ing o 34 co ical e-
gions (in igh and le hemisphe es), 8 subco ical egions segmen ed
om F eesu e , and he ce ebellum (Figu e 2.6, hea maps). Se -
e al o ou modules exhibi ed s ong o e lapping wi hin he RSNs.
Fo ins ance, module M6 encompassed he Soma o-mo o Ne wo k
(SMN) while also in eg a ed pa s o he Do sal and Ven al A en-
ion Ne wo ks (DAN, VAN). Module M18 showed an o e lap wi h a
po ion o he De aul Mode Ne wo k (DMN), and module M14 did i
wi h he Visual Ne wo k (VIS). Mo eo e , ce ain modules displayed
Resul s 35
high o e lapping wi h speci ic ana omical egions. Fo example, mod-
ule M20 included he medial and la e al O bi o on al co ices, while
module M26 in eg a ed he Basal Ganglia and Thalamus.
Figu e 2.5: S a is ical dependencies be ween di e en mul i-scale me ics. C oss-
co ela ion plo s be ween mul i-scale index (MSI), in a-s eng h (as a p oxy o module se-
g ega ion), size and heigh . Di e en colo -poin s ep esen di e en modules o he op imal
b ain pa cella ion wi h a o al numbe o 26 modules. P incipal-diagonal plo s show p obabil-
i y densi y unc ions o he di e en me ics. No ice ha highe co ela ions be ween module
size, module heigh , and in a-s eng h indica e ha heigh ened connec i i y wi hin modules
(seg ega ion) esul s in la ge dend og am modules and delayed mic o- egion spli ing in he
ee. The MSI e eals a dis inc ou lie (M1) encompassing se e al s uc u es such as a pa o
he p ecuneus, is hmus cingula e, and pos e io cingula e, loca ed a he bo de be ween DMN,
on o-pa ie al, and do sal a en ion ne wo ks.

Chap e 3
T ansc ip ional signa u es o
synap ic esicle genes de ine
myo onic dys ophy ype I
neu odegene a ion
In p e ious chap e s, I ha e desc ibed how Image De i ed Pheno-
ypes (IDPs) can be in e sec ed wi h he neu ogene ic da a om he
Allen Human B ain A las (AHBA) o s udy he neu ogene ic basis
o b ain- ela ed diso de s. In his chap e , we applied hese ech-
niques o desc ibe he genes associa ed o wo IDPs, olume loss
(VL) and neu opsychological de ici s (NDs) in pa ien s diagnosed
wi h myo onic dys ophy ype I (DM1), a monogenic (o Mendelian)
diso de .
3.1 In oduc ion
DM1 is a complex mul isys em disease ha a ec s skele al muscles
[134], hea [135], lungs [136], endoc ine sys em [137], egula ion o
sleep cycles [138], and o he aspec s o b ain ac i i y [139]. Epidemi-
ologically, DM1 is he mos common adul -onse muscula dys ophy
in humans, wi h a epo ed p e alence o 1/7,400 people wo ldwide
43
44 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
[140] and is abou h ee imes highe in Gipuzkoa [53], No he n
Spain (whe e his s udy was pe o med). Neu oimaging s udies ha e
shown he b ain damage in DM1 pa ien s, including g ey ma e
(GM) a ophy mainly a ec ing he on al and pa ie al lobes [141]
bu also, in he hippocampus [142] and o he subco ical s uc u es
[143]. Whi e ma e ac al e a ions (WM-TAs) ha e been widely
epo ed in DM1, in bo h c oss-sec ional [144, 145, 146, 147, 148,
149, 150, 151, 152] and longi udinal s udies [153]. These indica e
widesp ead WM-TA h oughou he whole b ain, bu ha is mo e
se e e in he on al, empo al and subco ical ac s. In ela ion
o neu opsychological pe o mance, execu i e unc ion, a en ion and
isuocons uc ion ha e been shown signi ican associa ion wi h WM-
TA [145, 154]. To he da e, he neu ogene ic p o iles o such pa e ns
o WM-TA o he ones co esponding o GM damage ha e no been
ye de e mined.
Using an app oach ha combines magne ic esonance imaging
(MRI) and la ge-scale b ain ansc ip omics, we aimed he e o assess
o wha ex en he s uc u al damage in he DM1 b ain ep esen s a
neu ogene ic signa u e. In con as o o he neu odegene a i e dis-
eases in which a la ge numbe o candida e genes a e implica ed, o
example, abou 700 genes in Alzheime ’s disease (AD) [155], DM1
is a monogenic diso de caused by a mu a ion in he gene encod-
ing he myo onic dys ophy p o ein kinase (DMPK) [156]. How-
e e , al hough he disease is monogenic, i s pheno ype is mainly due
o an abno mal ac i i y o he ibonucleic acid (RNA)-binding p o-
ein muscleblind-like 1 and 2 genes (MBNL1,MBNL2) [157, 158]
and CUGBP which egula es he exp ession o many o he genes,
such as he chlo ide channel 1 gene (CLCN1) ha egula es chlo ide
conduc ance du ing muscle de elopmen [159], he insulin ecep o
gene (INSR) [160], he b idging in eg a o 1 gene (BIN1) [161] o
o he genes di ec ly ela ed o he main symp oms o he condi ion
[162]. The neu opsychological p o ile o DM1 is mainly associa ed
wi h cogni i e di icul ies including isuospa ial p ocessing and exec-
u i e unc ioning [163, 164]. S ikingly, he DMPK pa hogenic geno-
ype has also been associa ed wi h o he genes ha encode p o eins
Me hods 45
implica ed in b ain neu odegene a ion, majo ly o au deposi s [165],
bu also amyloid be a (Aβ) [166] o alpha synuclein [167, 168]. Thus,
i is suspec ed ha he NDs and b ain damage ound in DM1 pa ien s
migh p esen some simila i ies o hose in o he neu odegene a i e
diseases, al hough his issue has ye o be ully add essed. The e o e,
despi e he monogenic o igin o DM1, he gene- o-gene in e ac ome
scales up o implica e mul iple sys ems in he b ain and body. To
da e, a p ecise associa ion be ween he en i e ansc ip ome and he
b ain neu odegene a ion and cogni i e de e io a ion in DM1 pa ien s
emains unexplo ed.
Some s udies ha e assessed he ela ionship be ween gene ics and
s uc u al b ain al e a ions in DM1, con i ming ha he numbe o
pa hogenic epea s o he cy osine- hymine-guanine (CTG) iple in
he DMPK gene (a pa ame e used o quan i y he molecula se e i y
o he disease) was associa ed wi h bo h mo e GM and whi e ma -
e a ophy, and o he NDs o hese pa ien s [147, 169]. He e, we
pe o med an in e sec ion analysis o neu oimaging pheno ypes and
he AHBA o la ge-scale ansc ip ional human da a [19], ollowing a
simila me hodology o ha used p e iously [32, 28, 34]. Ou hypo-
hesis was ha iden i ying he genes whose exp ession coincided mo e
closely wi h he b ain damage ound in DM1 pa ien s, we migh be e
unde s and he gene ela ionships associa ed wi h he b ain damage
ha a ises in his pa hology. Simila ly, we also assessed he ela ion-
ship be ween gene ansc ip ion in speci ic ana omical egions and
he b ain maps o NDs in hese pa ien s.
3.2 Me hods
3.2.1 Pa icipan s, wo coho s
A o al o N=95 subjec s we e ec ui ed o a c oss-sec ional s udy,
35 o whom we e DM1 pa ien s who we e ea ed a he Neu ology
Depa men o he Donos ia Uni e si y Hospi al (Gipuzkoa, Spain),
while 60 subjec s pa icipa ed as Heal hy con ols (HCs). All pa ien s
and HCs we e ec ui ed om he icini y o he Donos ia Uni e si y
46 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
Hospi al, and he wo g oups we e ma ched o age, gende and educa-
ion. The imaging da a om he DM1 pa ien s and HCs was acqui ed
a wo di e en Ins i u ions. A one, 19 DM1 pa ien s (mean age 53.3
yea s [SD ±8.1 yea s]; 9 males, 10 emales) and 29 HC (52.2 [±8.1]
yea s; 12 males, 17 emales) we e examined and a he second, 16
DM1 pa ien s (48.8 [±7.7] yea s; 7 males, 9 emales) and 31 HC (47.6
[±7.6] yea s; 14 males, 17 emales). Fo he mean alues, and he
compa isons be ween g oups and coho s see Tables 3.1 and 3.2.
Fi s
coho
Second
coho p E ec size
(Cohen)
Demog aphic
N19 16
Age, yea s 53.30 (8.09) 48.75 (7.72) 1.69 0.10 0.57
Males, n (%) 9 (47.37) 7 (43.75) 0.05* 0.83 -0.07
Educa ion, yea s 16.37 (4.87) 13.75 (4.85) 1.59 0.12 0.54
Clinical
N19 16
CTG expansion size 522.79 (448.44) 827.13 (433.08) -2.03 0.05 -0.69
MIRS 2.47 (0.96) 3.53 (0.83) -3.37 0.002 -1.17
Time be ween MRI and
CTG measu es, yea s 0.14 (0.14) 9.40 (7.23) -5.6 0.000 -1.93
Neu opsychological
N18 16
Visuospa ial -0.35 (1.21) -1.17 (1.34) 1.86 0.07 0.64
Ve bal memo y -0.04 (2.33) -0.74 (1.99) 0.94 0.36 0.32
A en ion -2.50 (1.92) -2.62 (2.14) 0.19 0.85 0.06
Execu i e unc ioning -0.91 (2.08) -1.84 (2.40) 1.21 0.23 0.42
Visual memo y -0.34 (0.97) -1.01 (1.14) 1.83 0.08 0.64
In elligence (IQ) 101.11 (11.51) 91.19 (13.61) 2.3 0.03 0.79
Table 3.1: Demog aphic, clinical and neu opsychological a iables. Mean alues
(s anda d de ia ions in b acke s) o he di e en a iables sepa a ed by coho . The neu o-
psychological a iables coincide wi h he di e en composi e-sco es om he di e en cogni i e
domains. All he neu opsychological a iables we e calcula ed using no ma i e da a om a
heal hy Spanish popula ion. All domains a e ep esen ed in Z sco es (Mean=0; SD=1) excep
IQ (Mean=100; SD=15). Smalle alues indica e wo se pe o mance. Bold alues indica e p <
0.05. *Fo gende di e ences, he χ2 es was used.
The DM1 pa ien s we e only included i hey had molecula con-
i ma ion o hei DM1 diagnosis, indica ing he expansion om i y
o housands o CTG epea s in he DMPK gene [156]. The diagnosis
was ob ained when pa ien s we e be ween 18 and 40 yea s old, and
he e o e ca ego ized as adul -onse DM1, as p oposed by he ou h
Me hods 47
DM1 HC p E ec size
(Cohen)
1s Coho
N19 29
Age, yea s 53.30 (8.09) 52.17 (8.05) 0.47 0.64 0.14
Males, n(%) 9 (47.37) 12 (41.38) 0.17* 0.68 -0.12
Educa ion, yea s 16.37 (4.87) 14.79 (3.36) 1.33 0.19 0.39
2nd Coho
N16 31
Age, yea s 48.75 (7.72) 47.55 (7.54) 0.51 0.61 0.16
Males, n(%) 7 (43.75) 14 (45.16) 0.01* 0.93 -0.03
Educa ion (yea s) 13.75 (4.85) - - - -
Table 3.2: Age, gende and yea s o educa ion be ween DM1 pa ien s and HCs.
Mean alues o di e en a iables a e shown (s anda d de ia ions a e gi en in b acke s). *Fo
gende di e ences, he Chi2 es was used.
Ou come Measu e o Myo onic Dys ophy Type 1 (OMMYD-4). Pa-
ien s we e excluded i a leas one o he ollowing c i e ia was me :
his o y o a majo psychia ic o soma ic diso de in acco dance wi h
Diagnos ic and S a is ical Manual o Men al Diso de s, i h edi ion
(DSM-V) c i e ia; acqui ed b ain damage; alcohol o d ug abuse; he
p esence o co po al pa amagne ic de ices like pacemake s o me al
p os hesis ha migh comp omise he MRI s udies; and he p es-
ence o b ain abno mali ies ha could a ec he olume ic analysis.
HCs sa is ied he same exclusion c i e ia bu he numbe o he CTG
epea s in hei DMPK gene anged om 5 o 34 [156]. DM1 pa-
ien s we e ec ui ed om he Neu omuscula Uni in he Neu ology
Depa men o he Hospi al Uni e si a io Donos ia, while HCs we e
ec ui ed om hei heal hy ela i es and gene al popula ion.
3.2.2 Demog aphic, clinical and neu opsycholo-
gical a iables
The demog aphic a iables o he subjec s eco ded we e hei age,
gende and yea s o educa ion. The clinical a iables we e he CTG
expansion size and Muscula Impai men Ra ing Scale (MIRS) sco e

48 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
[170]. Neu opsychological a iables co esponded o composi e al-
ues om di e en cogni i e domains ob ained h ough a comp e-
hensi e neu opsychological e alua ion pe o med by an expe ienced
neu opsychologis who was blind o he pa ien ’s clinical condi ion
(CTG expansion size and MIRS esul s). The neu opsychological
assessmen included se e al sub es s om he Wechsle Adul In-
elligence Scale III (WAIS III) [171], including: Digi span, Vocab-
ula y, Block design, Objec assembly, A i hme ic, and Simila i ies.
O he cogni i e es s used we e: S oop es , Cali o nia Compu e -
ized Assessmen Package (CALCAP), Ra en’s p og essi e ma ices,
Rey Audi o y Ve bal Lea ning Tes (RAVLT) [172], Wo d Fluency
[173, 174], Rey-Os e ie h Complex Figu e es (ROCF) [175] and
Ben on’s Judgemen o Line O ien a ion [176]. The pa ien s’ aw
sco es we e con e ed in o s anda dized - alues based on he no m-
a i e sco es o he Spanish popula ion in each es . Finally, he
di e en neu opsychological sco es we e educed in o six di e en do-
mains: isuospa ial (Block design and ROCF copy), e bal memo y
(RAVLT immedia e ecall, RAVLT delayed ecall, To al RAVLT),
a en ion (Digi span, STROOP wo d, STROOP colou , Simple Re-
ac ion Time (RT), elec ion RT, Sequen ial 1 RT, Sequen ial 2 RT),
execu i e unc ioning (To al RAVEN, seman ic luency, phonemic lu-
ency, STROOP colou -wo d, STROOP in e e ence), isual memo y
(ROCF delayed ecall), and in elligence (es ima ion o IQ was based
on Vocabula y, Block design, Objec assembly, A i hme ic, and Sim-
ila i ies sub es s).
3.2.3 MRI acquisi ion and p e-p ocessing
Fo he i s coho , MRI was conduc ed on a 3 Tesla scanne (T i-
oTim, Siemens) using a high- esolu ion 3D sequence o magne iza ion-
p epa ed apid acquisi ion wi h g adien echo (MPRAGE) and apply-
ing he ollowing pa ame e s: Sagi al 3D T1 weigh ed acquisi ion,
epe i ion ime (TR) = 2,300 ms, echo ime (TE) = 2.86 ms, in e -
sion ime = 900 ms, lip angle = 9°, ma ix = 192 ×192 mm2, slice
hickness = 1.25 mm, oxel dimensions = 1.25 ×1.25 ×1.25 mm3,
numbe o signals a e aged (NSA) = 1, slices = 144, no gap, o al
Me hods 49
scan du a ion = 7 min and 22 s. Di usion-weigh ed images we e ac-
qui ed using an echo plana imaging (EPI) sequence: 1.75 ×1.75 ×
2 mm oxels; 77 axial slices; b alue o 1,000 s/mm2; 64 di ec ion
di usion-weigh ed and 1 baseline image; TR = 10,000 ms, TE = 92
ms; angle 90, acquisi ion ma ix size = 122 ×122.
Fo he second coho , MRI was conduc ed on a 1.5 Tesla scan-
ne (Achie a No a, Philips), using a high- esolu ion olume ic u bo
ield echo (TFE) sequence wi h he ollowing pa ame e s: Sagi al 3D
T1 weigh ed acquisi ion, TR = 7.2 ms, TE = 3.3 ms, in e sion ime
= 0 ms, lip angle = 8°, ma ix = 256 ×232 mm2, slice hickness =
1 mm, oxel dimensions = 1 ×1×1 mm3, NSA = 1, slices = 160,
no gap, o al scan du a ion = 5 min and 34 s. Di usion-weigh ed
images we e acqui ed using a single sho spec al p esa u a ion wi h
in e sion eco e y (SPIR): 1.75 ×1.75 ×2 mm oxels; 60 axial slices;
b alue o 800 s/mm2; 32 di ec ion di usion-weigh ed and 1 baseline
image; TR = 9,967 ms, TE = 66 ms; angle 90, acquisi ion ma ix size
= 128 ×128.
To pe o m oxel-based GM mo phome ic compa isons be ween
he subjec s, DM1 and HCs, we pe o med oxel-based mo phome y
(VBM) ollowing a p ocedu e simila o ha used p e iously [4, 177],
an op imized VBM p o ocol [178] ca ied ou wi h he FSL 6.01 so -
wa e. Fi s , skull emo al was pe o med, ollowed by GM segmen a-
ion and egis a ion o he MNI 152 s anda d space using non-linea
egis a ion [179]. The esul ing images we e a e aged and lipped
along he x-axis o c ea e a le - igh symme ic, s udy-speci ic GM
empla e. Second, all na i e GM images we e non-linea ly egis e ed
o his s udy-speci ic empla e and ‘modula ed’ o co ec o local ex-
pansion (o con ac ion) due o he non-linea componen o he spa-
ial ans o ma ion. The modula ed GM images we e hen smoo hed
wi h an iso opic Gaussian ke nel a sigma = 3, and inally, he pa -
ial GM olume es ima es no malized o he subjec ’s head size we e
compa ed.
WM-TA we e assessed wi h ac -based spa ial s a is ics (TBSS)
om FSL [180] using ac ional aniso opy (FA) images. Fi s , all
indi idual FA images we e no malized o a common empla e using
50 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
a non-linea ans o ma ion. Nex , he mean image ac oss all sub-
jec s was compu ed and skele onized o ge he mean FA skele ons,
which ep esen egions wi h high con idence bundles common o all
subjec s, hus emo ing some o he indi idual subjec -speci ic ac -
based he e ogenei ies. FA images o each subjec we e hen p ojec ed
on o he mean skele on.
3.2.4 Imaging s a is ical analysis
Fo he VBM analyses, a gene alized linea model was i ed o each
oxel and image using he FSL so wa e, con olling o age and head
size wi h wo di e en con as s: DM1 <HC and DM1 >HC. All
he esul s we e ob ained wi h wo- ailed es s, co ec ing o mul iple
compa isons using he Mon e Ca lo simula ion clus e -wise co ec ion
implemen ed in he AFNI 19.3.00 so wa e, and using 10,000 i e -
a ions o es ima e he p obabili y o alse posi i e clus e s wi h a p
alue <0.05. Each coho was analysed sepa a ely and in combin-
a ion, as explained below, al hough s a is ical compa isons be ween
he wo coho s we e no pe o med.
Fo he TBSS analyses, g oup compa ison was pe o med us-
ing he andomise ool in FSL, a non-pa ame ic pe mu a ion es
o inding signi ican s a is ical di e ences be ween g oups a he
oxel le el. Fo mul iple compa ison co ec ion, we used h eshold-
ee clus e enhancemen (TFCE)[181] wi h a numbe o i e a ions o
5,000 and amily-wise e o (FWE) co ec ed p= 0.05, hus ensu ing
ha he chance o alse posi i es is no mo e han 5%, o equi alen ly,
ensu ing 95% con idence o no alse posi i es. G oup compa ison was
pe o med using wo di e en con as s, DM1 >HC and HC >DM1.
3.2.5 T ansc ip omics b ain maps
To build b ain maps o ansc ip ion, we ook ad an age o he pub-
licly a ailable da a om he AHBA (h p://human.b ain-map.o g/) [19].
The da ase consis ed o MRI images and a o al o 58,692 mic oa ay-
based ansc ip ion p o iles o abou 20,945 genes sampled o e 3,702
di e en egions ac oss he b ains o six humans. O he o al 3,702
Me hods 51
sampling si es, 2,728 we e loca ed in co ical and subco ical GM, 368
in he ce ebellum, 586 in he b ain s em, and 15 in whi e ma e . Ou
analyses he e a e es ic ed o he 2,728 si es co e ing he co ical and
subco ical GM. To pool all he ansc ip ion da a in o a single b ain
empla e, we ollowed a simila p ocedu e o ha employed elsewhe e
[25]. Fi s , o e-anno a e he p obes o genes we made use o he
e-anno a o oolki [182]. Second, we emo ed hose p obes wi h
insu icien signal by looking a he sampling p opo ion (SP), which
was calcula ed o each b ain as he a io be ween he samples wi h a
signal g ea e han he backg ound noise di ided by he o al numbe
o samples. P obes wi h a SP lowe han 70% in any o he six b ains
we e emo ed om he analysis, he eby ensu ing su icien sampling
powe in all he b ains. A e ha , we chose he alue o he p obe
o each gene wi h he maximum di e en ial s abili y (DS), accoun -
ing o he ep oducibili y o gene exp ession ac oss b ain egions
and indi iduals, and calcula ed using spa ial co ela ions simila o
hose employed p e iously [26]. Fo his he au oma ed ana omical
labelling (AAL) a las was used [84] om which he ce ebellum was
excluded, esul ing in 90 di e en ana omical egions 1. Finally, o
emo e he in e subjec di e ences, he ansc ip ion alues o each
gene and b ain we e ans o med in o Zsco es and pooled oge he
om he six di e en b ains, ob aining a single map using he MNI
coo dina es p o ided in he da ase . Finally, o elimina e he spa ial
dependencies o he ansc ip ion alues a he sampling si es ( ha
is, o co ec o he ac ha nea es si es ha e mo e co ela ed an-
sc ip ion), we inally ob ained a single ansc ip ion alue o each
egion in he AAL a las by calcula ing he median o all he alues
belonging o he gi en egion.
1AAL egions we e e oded wi h a Gaussian ke nel wi h a ull wid h a hal
maximum (FWHM) equal o 2 mm, he eby elimina ing alse-posi i e sampling
si es, i.e.: hose ha do no belong o he egion o in e es bu o one in he
neighbo hood.
58 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
Figu e 3.1: Me hodological scheme o he associa ion be ween ansc ip omics and
a ophy in DM1, measu ed as b ain VL.A: Two coho s o DM1 pa ien s we e ec ui ed
(o ange and blue) and we ob ained he b ain maps o he VL o each, compa ing he images
wi h a g oup o HCs using VBM and TBSS, co ec ing o mul iple compa isons. We aimed
o cha ac e ize he associa ion be ween VL and ansc ip omics, assessing he simila i y in he
spa ial pa e ns o VL ac oss b ain egions and he spa ial pa e ns o gene ansc ip ion om
he AHBA da ase , p e-p ocessed ollowing a pipeline ha is summa ized in six main s eps
( o u he de ails, see Me hods). A e unning he AHBA pipeline, abou 14 K genes inally
had ansc ip ion alues used in he analysis om he 58-K p obes o iginally a ailable. The
ed egions in he b ain co espond o he si es a which ansc ip ion was sampled. *O he
o al 3,702 sampling si es, 2,728 we e loca ed in co ical and subco ical g ey ma e , 368 in
he ce ebellum, 586 in he b ain s em, and 15 in whi e ma e . Ou analyses he e a e es ic ed
o he 2,728 si es co e ing he co ical and subco ical g ey ma e . B: SP and DS o he 27
p eselec ed hypo hesis-d i en genes included in he lis o candida es ele an o DM1 (ob ained
by e iewing he li e a u e). The CACNA1S,CLCN1,KL and SIX5 genes did no ha e a mean
SP alue abo e 70% and hus, hey we e excluded om u he analyses. O he emaining 23
genes, he maximum DS co esponded o SNCA and he minimum DS o DMPK (see a ows).
By examining he spa ial simila i y in he ansc ip ion alues, he emaining 23 candida e genes
we e clus e ed in o h ee g oups. The blue one o med by DMD,SNCA and MAPT played a
majo ole in he cha ac e iza ion o VL.

Resul s 59
A lis o he 27 mos ele an hypo hesis-d i en genes in DM1 was
d awn up (Figu e 3.1, panel B; o de ails on e e ences suppo ing he
selec ion o each gene, see Table 3.3). O hese 27 p eselec ed genes,
ou genes we e disca ded based on hei SP and DS alues (KL, SIX5,
CLCN1, CACNA1S). These 4 genes had less han 70% SP (Figu e
3.1, panel B), which implied insu icien ansc ip ion signal ac oss
all he sampling si es. The e o e, he inal lis o he mos ele an
hypo hesis-d i en genes con ained: DMPK, HSPB2, INSR, CPEB4,
ANK2, ARHGEF7, SOS1, PHKA1, MBNL1, KIF13A, APP, MAPT,
SNCA, MBNL2, RIPK1, PRNP, TARDBP, MAP3 K7, BIN1, DMD,
LDB3, TTN and CAPN3. When he pai wise gene- o-gene simila -
i y in ansc ip ion was e alua ed ac oss he b ain o hese 23 genes,
Silhoue e maximiza ion iden i ied h ee clus e s (in yellow, blue and
g een in Figu e 3.1, panel B), indica i e o unc ional simila i ies in
he ansc ip ion signals among he 23 mos ele an genes.
We nex applied a DDS ha in ol ed iden i ying he genes om
he en i e ansc ip ome wi h maximal associa ion in he VL maps,
esul ing in a o al o 370 N genes and 187 P genes om he i s
coho and 441 N-genes and 161 NP-genes om he second one. The
genes common o he wo coho s we e hose inally used in he ana-
lysis, a o al o 251 N genes and 101 P genes (Figu e 3.2, panel A).
In e es ingly, wo o he genes in he lis o he 23 mos ele an
hypo hesis-d i en genes also appea ed in he lis o N genes, SNCA
and DMD, displaying a simila ansc ip ion pa e n as he MAPT
gene (blue clus e in Figu e 3.1, panel B). The spa ial co ela ion o
SNCA and DMD ansc ip ion wi h he amoun o VL in he di e -
en b ain egions p o ed o be nega i e o he wo genes and smalle
in he wo coho s: <-0.52 o DMD and <-0.70 o SNCA (Fig-
u e 3.2, panel B). In addi ion o iden i ying he N and P genes, we
also sea ched o he NC and PC genes (Figu e 3.3, panel A) ha
ep esen hubs owa ds he N and P ails o he exp ession simila i y
ma ix. We ound 452 NC genes, 396 PC genes and 238 genes con-
nec ing bo h N and P ails. Rema kably, in he g oup o PC genes, we
ound LDB3,CAPN3 and HSPB2 ha we e in he lis o hypo hesis-
d i en genes, belonging o he same clus e o exp ession simila i y
60 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
(g een clus e , Figu e 3.1, panel B). In addi ion, he p eselec ed gene
PHKA1 was also common o he g oups o NC and PC genes.
Figu e 3.2: Da a-d i en s a egy o de e mine he associa ion be ween he an-
sc ip ome and VL in DM1.A: His og am o he spa ial-co ela ion alues (measu ed as he
Z-sco e) be ween VL and ansc ip ional ac i i y o all he genes in bo h coho s. Fo bo h
coho s he N genes (Z<-2) and P genes (Z>2) a e colou ed in blue and ed, espec i ely.
The inal lis o genes used o u he analyses a e hose ha a e common o he wo coho s,
consis ing o 251 N genes and 101 P genes. F om all he genes ha p o ide a maximum asso-
cia ion be ween VL and ansc ip omics, only wo genes we e in he panel o p eselec ed genes:
SNCA and DMD.B: B ain maps o ansc ip ion in he b ain egions o he wo genes DMD
and SNCA, which p o ided a high spa ial co ela ion ( ) wi h he VL b ain maps o bo h
coho s.
Pooling he N, P, NC and PC genes oge he , along wi h hose
common o bo h he NC and PC ca ego ies, we adop ed a DDS o
achie e unsupe ised clus e ing o he exp ession simila i y ma ix,
iden i ying wo majo clus e s a e Silhoue e maximiza ion (in blue
and ed in Figu e 3.3, panel B). Impo an ly, he N and P genes
ully seg ega ed in o he wo di e en ia ed clus e s, wi h all he N
genes belonging o he blue clus e and all he P genes o he ed
one, he eby con i ming he di e en unc ional oles o he g oups o
genes in he N and P ails. These wo clus e s we e used sepa a ely
o gene en ichmen analysis. The sea ch o he GO biological p o-
Resul s 61
Figu e 3.3: Func ional desc ip ion o he genes wi h he highes associa ion wi h
VL.A: An all- o-all gene-exp ession simila i y ma ix iden i ied he connec o hub genes. A
o al o 1,086 genes we e ound, equal o he sum o he NC = 452 (blue), PC = 396 ( ed) and
238 common NC and PC genes (g ey). B: Two clus e s we e inally ound ha pooled all gene
classes, he blue one con ains he o iginal N = 251 genes and he ed one con aining he o iginal
p = 101 genes. C: Gene en ichmen o GO biological p ocess and Reac ome pa hways: in blue
a e he neg-co genes and in ed, he pos-co genes. Abb e ia ions: Ac ., ac i a ion; CNS,
cen al ne ous sys em; Mod., modula ion; Neg., nega i e; Pos., posi i e; Rec., ecep o /s; Reg.,
egula ion
cesses and Reac ome pa hways con i med he di e en ia ed oles o
hese wo clus e s, wi h he neg-co genes mo e ela ed o neu onal
and synap ic unc ion, in ol ing key synap ic esicle e en s such as
ecycling, localiza ion, endocy osis and exocy osis bu also, he dy-
namics o se o onin and dopamine neu o ansmi e elease (Figu e
3.3, panel C). By con as , he clus e o pos-co genes was mo e
ela ed o non-neu onal ac i i ies, such as in e e on signalling, en-
do helial cell di e en ia ion, angiogenesis, blood anspo and cell
de elopmen .
To assess how he ansc ip omics co ela ed wi h ND, we ocused
on he neu opsychological domains in which he composi e sco e e-
62 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
lec ed s ong impai men , sa is ying Z<-2, which was only he case
o he a en ion ca ego y ( i s coho Z= -2.5, second coho Z= -
2.62: see Table 1 and Figu e 3.4, panel A o he dis ibu ion o all he
Z-sco e alues om he wo coho s). As indica ed in he Me hods,
he a en ion composi e was ob ained by a e aging he Z-sco es o
he ollowing domains: digi span, STROOP wo d, STROOP colou ,
simple RT, elec ion RT, sequen ial 1 RT and sequen ial 2 RT.
Figu e 3.4: Da a-d i en s a egy o de ine he associa ion be ween he ansc ip-
ome and a en ion co-ac i a ion maps in DM1.A: A en ion sco es measu ed as Z-
sco es o he wo coho s. Because he Z-sco es we e no malized o he alues in he HCs,
nega i e alues o Zindica e wo se pe o mance han he HCs. B: A en ion co-ac i a ion
maps buil wi h he Ginge ALE ool and p ojec ed on o he a las. C: His og ams o he spa ial
co ela ions be ween he Z-sco es o he a en ion maps and he ansc ip ional ac i i y o each
gene. The ail o N genes (Z<-2, colou ed in blue) includes he SNCA gene om he lis o
p eselec ed genes, whe eas he ail o he P genes (Z>2, ed) does no include any o hese.
Following a p ocedu e simila o ha desc ibed in Figu e 3.3 (panel A), we iden i ied he PC
genes ( ed), NC genes (blue, including MAPT), and hose common o he NC and PC (g ey,
including PHKA1). D: A e pooling all classes o genes oge he and clus e ing, wo g oups
we e de ined: one including all he neg-co genes (blue, wi h SNCA and MAPT) and one wi h
he pos-co genes ( ed, wi h PHKA1 ). Gene en ichmen o he ags GO biological p ocess and
Reac ome pa hways. As in Figu e 3.3 (panel C), he wo clus e s also ep esen ed wo sepa a ed
unc ions: he neg-co one co ela ed wi h neu onal unc ions, while he pos-co co ela ed o
non-b ain unc ions. Abb e ia ions: Mod., Modula ion; Neg., Nega i ek; Pos., Posi i e; Reg.,
Regula ion
Resul s 63
We hen ob ained b ain co-ac i a ion maps using ‘A en ion’ as
a keywo d o he sea ch in Ginge Ale (Figu e 3.4, panel B). As o
VL, we calcula ed he associa ion be ween he a en ion co-ac i a ion
maps and ansc ip omics by calcula ing he spa ial Pea son co ela-
ion be ween he wo Z sco e ec o s (one alue pe b ain egion in he
a las, see he his og am o all he co ela ions in Figu e 3.4, panel C).
The e was a o al o 217 N genes (including SNCA), 54 P genes, 336
NC genes (including MAPT om he lis o p eselec ed hypo hesis-
d i en genes), 265 PC genes, and 156 genes common o bo h he NC
and PC gene se s (including PHKA1). Pooling oge he all classes
o genes, we ollowed a DDS o unsupe ised clus e ing and iden i-
ied wo clus e s a e Silhoue e maximiza ion (Figu e 3.4, panel C).
Like VL, he wo clus e s we e highly seg ega ed and inco po a ed
all neg-co genes in one clus e , which was en iched in genes ela ed
o neu onal and synap ic unc ion (Figu e 3.4, panel D), wi h all
he pos-co genes in he o he clus e en iched in non-b ain- ela ed
ac i i ies.
To alida e some o he p edic ions done by ou compu a ional
s a egy linking neu odegene a ion and neu opsychology wi h an-
sc ip omics, we also pe o med neu opa hological analyses o b ain
samples om DM1 pa ien s. Neu opa hological de ails o he cases
a e shown in Table 3.4. O e all, mode a e ce eb al a ophy wi h en-
la ged en icles was ound in e e y DM1 case. The main common mi-
c oscopical al e a ions we e p e- angles, neu o ib illa y angles, and
neu opil h eads in selec ed egions o he b ain s em and ce eb um.
NFTs we e s ained wi h an ibodies AT8 (P- au Se 202/Th 205), P-
au ThR181, P- au Se 422, 3R au, and 4R au. The dis ibu ion was
a iable depending on he case. Cases I09-254 and I13-132 we e ca -
ego ized as s ages I/II o B aak; cases I09–255, I13–131, and I13–134
as s ages III/IV; and cases I09–256 and I13–133 as s ages V/VI. β-
amyloid deposi ion was absen in i e cases, and p esen in cases
I13–134 and I13–133 co esponding o Thal phases 1 and 3, espec -
i ely. Mild as ocy ic gliosis and mic ogliosis we e obse ed in he
same egions wi h NFTs. Dys ophic neu i es, eac i e as ocy es
and ac i e mic oglia we e ound a ound β-amyloid deposi s in senile

64 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
plaques. Lewy bodies we e ound in one case (I13–133) co espond-
ing o s age 3 o B aak o Pa kinson’s pa hology. TDP-43 p o eino-
pa hy was absen in e e y case. Mild o mode a e small blood essel
disease, cha ac e ized by a he ioloscle osis and a e iola hyalinosis,
was common. This was accompanied by mild s a us c ibosus in he
s ia um, halamus and whi e ma e in ou cases. The whi e ma -
e o he ce eb um was educed in size and showed mild o mod-
e a e myelin pallo in i e cases (I09–255, I13–131, I13–132, I13–133
and I13–134); he co pus callosum was hinne . Myelin changes, as
seen wi h Kl¨u e –Ba e a we e accompanied by loss o ne e ib es as
isualized wi h an i-neu o ilamen an ibodies. Axonal ballooning was
absen . No in lamma o y changes we e obse ed in he whi e ma e ,
wi h he excep ion o a ew pigmen -laden pe i ascula mac ophages.
Co po a amylacea we e seldom obse ed in he sub en icula egion,
pe i ascula spaces and subpial pa enchyma. One case had su e ed
om a ecen in a c ion in he le occipi al co ex. Th ee cases had
mild Pu kinje cell loss (I13–131, I13–132 and I13–133), and wo had
mild a ophy o he in e io oli es accompanied by as ocy ic gliosis;
ep esen a i e au pa hology in Figu e 3.5 lowe panel.
case age NFT
B aak SP Thal LB
B aak TDP-43 o he s Wes e n blo ing P- au-Se 214,
au bands
I09–
254 60 S age
II No No No — 58/59 kDa,
I09–
255 58 S age
IV No No No sb d, sc, wmd,
in oli 58 kDa, 55 kDa
I09–
256 74 S age
VNo No No in oli 58 kDa, 55 kDa
I13–
131 67 S age
III/IV No No No sb d, sc, wmd,
Pc↓
weak 68 kDa, 64 kDa, 60 kDa,
58 kDa, 55 kDa
I13–
132 58 S age
I-II No No No sb d, wmd, Pc↓64 kDa, 60 kDa, 58 kDa, 55 kDa
I13–
133 68 S age
V/VI Thal 3 LB 3 No sb d, sc, wmd,
Pc↓
68 kDa, 64 kDa, 60 kDa, 58 kDa,
55 kDa, lowe bands
I13–
134 59 S age
III/IV Thal 1 No No sb d, sc, wmd,
in a c 66 kDa, 64 kDa, 58 kDa, 55 kDa
Table 3.4: Neu opa hological de ails o cases s udied.Abb e ia ions: in oli : mild o
mode a e ne e cell loss and mild as ocy osis in he in e io oli es; in a c : acu e occipi al
in a c ion; LB B aak: Lewy body pa hology acco ding o B aak s ages; NFT B aak: B aak
s ages o neu o ib illa y angle pa hology; Pc↓: mild o mode a e loss o Pu kinje cells in he
ce ebellum; sb d: small blood essel disease (a e ioloscle osis, a e iola hyalinosis); sc: s a us
c ibosus and dila a ion o pe i ascula spaces; SP Thal: phases o senile plaques acco ding o
Thal scale; TDP-43: TDP-43 p o einopa hy; wmd: mild o mode a e demyelina ion o he
ce eb al whi e ma e .
Resul s 65
Figu e 3.5: Neu opa hological analyses o b ain samples e ealed a high he e o-
geneous Tau-pa hology in DM1, di e en o he one in AD.Uppe panel: Gel elec-
opho esis and Wes e n blo ing o sa kosyl-insoluble ac ions o he hippocampus incuba ed
wi h an ibodies agains 3R au, 4R au, and phospho yla ed au a se ine 214 (P- au-Se 214)
in DM1 cases, and one AD o con ol. AD is cha ac e ized by h ee bands o 68, 64 and
60 kDa and an uppe weak band o 73 kDa. DM1 cases a e cha ac e ized by he p esence o
lowe bands o 58 and 55 kDa, and occasional uppe bands o 60 and 64 kDa; he band o
68/69 kDa is ba ely p esen . This pa icula pa e n, including di e ences om one case o
ano he , is associa ed wi h low molecula weigh 3R au and 4R au iso o ms sugges ing complex
al e ed Tau splicing. Lowe panel: Rep esen a i e images o NFTs and neu opil h eads in
he en o hinal co ex (EC), CA1 a ea o he hippocampus (CA1), empo al co ex (TC), and
pa ie al co ex (PC), isualized wi h he an ibody AT8, and an i-3R au and 4R au an ibodies.
Pa a in sec ions sligh ly coun e s ained wi h haema oxylin; ba =25 µm
Wes e n blo s o sa kosyl-insoluble ac ions s ained wi h an i-
Tau-P-Se 214 an ibodies e ealed h ee bands o 68 kDa, 64 kDa,
60 kDa and an uppe band o 73 kDa in one case wi h AD s age
V. This pa e n was in s iking con as wi h DM1 cases I09–254,
I09–255 and I09–256 p ocessed in he same memb ane ha appa -
en ly showed wo main bands o 58 and 55 kDa uni ed by a smea ,
and no bands o uppe molecula weigh s. The incuba ion o o he
memb anes wi h an i-4R au and an i-3R au showed simila al e a-
ions. Unique bands o low molecula be ween 50 and 60 kDa we e
66 Chap e 3. T ansc ip ional signa u es o DM1 neu odegene a ion
in s iking con as wi h he bands be ween 60 and 70 kDa obse ed
in AD. Nex , he se en cases wi h DM1 we e un in he same gel
and blo ed wi h an i-P- au-Se 214 an ibodies. Two bands o 58 and
55 kDa cha ac e ized cases I09–255 an I09–256; one band o abou
59/60 kDa cha ac e ized case I09–254. Howe e , ou bands o abou
64, 60, 58 and 55 kDa we e ob ained in cases I13–131, I13–132 and
I13–134. Finally, an addi ional band o abou 68 kDa and bands o
molecula weigh below 50 kDa we e iden i ied in case I13–133. In
ag eemen wi h he di e en phospho- au bands depic ed wi h he
an ibody P- au-Se 214, he blo s using an i-4R au and an i-3R au
equally disclosed indi idual pa e ns o bands pa icula ly linked o
3R au. Biochemical de ails o au bands in e e y case a e shown in
Table 3.4; ep esen a i e Wes e n blo s in Figu e 3.5 uppe panel.
We also s udied he exp ession o selec ed genes om hippocam-
pal issue om DM1 pa ien s. The s udy o messenge -RNA (mRNA)
le els o genes ela ed o synap ic esicles e ealed down- egula ed ex-
p ession o SNAP25 (p= 0.035) and a end o educed exp ession
o SYP and SYN1 (p= 0.06 and p= 0.08, espec i ely) be ween
DM1 and NFT cases. Rega ding he exp ession o genes ela ed o
s uc u al componen s o synapse, HOMER1 mRNA was signi ican ly
educed in NFT cases compa ed wi h MA (p= 0.037). PCLO and
ABLIM2 showed a endency o dec ease (p= 0.07 and p= 0.07,
espec i ely) in NFT cases compa ed wi h MA cases. None o his
g oup o genes was al e ed in DM1. The s udy o he exp ession o
GABAe gic- and glu ama e gic- ela ed genes showed signi ican ly e-
duced le els o GABRD ansc ip in NFT cases when compa ed wi h
MA (p= 0.05) and a endency o educed exp ession o GABRG2
in NFT cases compa ed wi h MA (p= 0.10) and DM1 cases (p=
0.09) (Figu e 3.6). Ve y s iking, om he genes analysed in Fig-
u e 3.6, genes SNCA, SNAP25, SYP, SYN1, SLC17A7 ( om syn-
ap ic esicles; panel A), FRMPD4, PCLO, BSN, ARPC5L ( om syn-
apse s uc u al componen s; panel B), and GABRA3, GRIA1 ( om
GABAe gic- and glu ama e gic- ela ed genes; panel C), all o hem
appea s in he neg-co gene lis used in he en ichmen ob ained
a e linking VL o ND wi h ansc ip omics. The endency showed
Resul s 67
in Figu e 3.6 o his se o genes is o be down egula ed wi h espec
o MA and NFL, in ag eemen wi h he nega i e-associa ion o hese
genes wi h VL.
Figu e 3.6: RNA exp ession in hippocampal issues om DM1 pa ien s e ealed
p o ein dys unc ion o synap ic esicle p ocesses in DM1. mRNA exp ession o synapse-
ela ed genes was analysed in he hippocampus in middle-aged (MA) con ol cases, myo onic
dys ophy 1 (DM1) cases, and cases wi h neu o ib illa y angles (NFT con ol) linked o AD-
ela ed pa hology a middle and ad anced s ages (III-VI) o B aak and B aak. A: Synap ic
esicles coding genes. B: Synapse s uc u al componen s coding genes. C: GABAe gic- and
glu ama e gic- ela ed coding genes. All da a we e exp essed as mean alues ±SEM. Di e ences
be ween g oups a e s a is ically signi ican a *p <0.05; ends a e indica ed wi h * and he
exac p alue
74 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
ypically de eloping con ol pa icipan s.
4.1 In oduc ion
Au ism encompasses mul iple mani es a ions, om impai ed social
communica ion and language o es ic ed o epe i i e beha io pa -
e ns, in e es s, and ac i i ies [239, 240, 241]. Due o he as he e o-
genei y in beha io , and as ecommended in DSM-5, his condi ion
is e e ed o as au ism spec um diso de (ASD), in which he e m
“spec um” emphasizes he a ia ion in he ype and se e i y o mani-
es a ions [242]. ASD is hough o esul om complex in e ac ions
du ing de elopmen be ween gene ic, cellula , ci cui , epigene ic, and
en i onmen al ac o s [243, 244, 245, 246, 247]. Se e al esea che s
ha e sugges ed ha an exci a ion/inhibi ion (E/I) imbalance du ing
de elopmen [248, 249] may be an essen ial mechanism, ye speci ic
ac o s d i ing he condi ion a e no well unde s ood. The apeu ic
in e en ions aiming o es o e he E/I balance in ASD a e a majo
challenge [250].
Conce ning neu obiology, he e ogenei y in b ain mo phology [251]
and b ain ne wo ks has been ound, e.g., in he on al, de aul mode,
and salience ne wo ks [252, 253, 254, 255, 256, 257], as well as in he
social ne wo k [258] — encompassing he p ima y mo o co ex, usi-
o m gy us, amygdala, ce ebellum, insula, soma osenso y co ex, and
an e io cingula e co ex [259, 255, 260]. ASD is also he e ogeneous
in ela ion o ne wo k cha ac e is ics; less seg ega ion and g ea e e -
iciency [261, 262], and he opposi e as well [263] o a combina ion o
bo h [264, 265], ha e been shown. Fu he mo e, ASD neu oana om-
ical co ela es a e no s a ic bu unde go changes h oughou de elop-
men [266, 5, 267], and he same seems o occu beha io ally in social
unc ioning and communica ion [268]. Al oge he , accumula ed e id-
ence has shown high he e ogenei y wi hin ASD in he pa icipa ion o
unc ional b ain ne wo ks and beha io al mani es a ions and in he
longi udinal ajec o ies a he indi idual le el.
Mo eo e , wi h neu oimaging s udies, ecen wo k has shown addi-
ional sou ces o he e ogenei y due o a ia ions in diagnos ic and in-

In oduc ion 75
clusion c i e ia and di e ences in he p ocessing neu oimaging pipeline
[269, 270]. ASD is also a polygenic, highly he e ogeneous condi ion,
wi h 1010 genes associa ed wi h ASD as o July 8, 2022, acco ding
o he Simons Founda ion Au ism Resea ch Ini ia i e (SFARI) gene
human da abase [see also h ps://gene.s a i.o g/da abase/human-gene/]. O
hose, 213 ha e a ele ance sco e o 1, meaning ha hey ha e max-
imum published pa hophysiological e idence o ASD. This high ge-
ne ic complexi y is ano he mani es a ion o he he e ogenei y o his
condi ion on se e al scales. P e ious wo k has assessed he associ-
a ions be ween ansc ip omics and b ain mo phology [225], showing
ha genes ha a e down egula ed and en iched o synap ic ans-
mission in indi iduals wi h au ism we e associa ed wi h a ia ions in
co ical hickness.
No el s a egies o ASD sub yping a e needed o o e come such
mul iscale he e ogenei y, which is he mos signi ican challenge in he
de elopmen o e ec i e he apies. Some s udies ha e add essed he
he e ogenei y in ASD o be e s a i y his condi ion [271, 272, 273,
274]. P e ious wo k pe o med clus e ing, pooling oge he ASD and
ypically de eloping con ol (TDC) g oups [271], and ound 2 g oups
o indi iduals showing hype connec ed o hypoconnec ed pa e ns
(each g oup con aining bo h ASD and TDC pa icipan s). S a i ic-
a ion yields educed in e indi idual di e ences and, he e o e, could
complemen —and e en alle ia e— he need o la ge sample sizes in
au ism-based bioma ke disco e y [275]. He e, and ollowing p e i-
ous wo k [276, 277, 273], we looked a la ge-scale b ain connec i i y
pa e ns common wi hin g oups o indi iduals o deploy sub yping
in ASD. In pa icula , we applied consensus clus e ing s a egies o
mul i a ia e connec i i y pa e ns o b ain egions [278, 279] o as-
socia ing connec i i y-based ASD sub ypes wi h hei neu ogene ic
p o ile. Following p e ious wo k [32, 28, 29, 223, 40, 222, 224, 34],
we hypo hesized di e en biological cha ac e iza ion unde lying he
neu ode elopmen al and ma u a ion b ain connec i i y p o ile o
each sub ype, unknown o his condi ion. Fo his, we used he
AHBA o whole-b ain ansc ip ional da a [19] and pe o med sub-
yping on 657 indi iduals wi h ASD om he Au ism B ain Ima-
76 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
ging Da a Exchange (ABIDE) eposi o y [280], all o hem ha ing
passed a e y s ic quali y assu ance c i e ion o elimina ion o pa -
icipan s by head mo emen du ing image acquisi ion, hus co ec -
ing a well-known spu ious excess o unc ional connec i i y d i en by
head mo emen s, which is e en mo e p onounced in he au is ic condi-
ion. Mo eo e , o o e come in e scanne a iabili y in he unc ional
connec i i y alues ac oss di e en ins i u ions, we applied igo ous
ha moniza ion s a egies o ans o m he e ogeneous da a in o equi-
alen s [281, 282, 283, 284].
4.2 Me hods
4.2.1 Pa icipan s
A o al o 2156 pa icipan s om he ABIDE-I [280] and ABIDE-
II [285] eposi o ies we e ini ially conside ed in his s udy, o which
1026 we e indi iduals wi h ASD and 1130 we e TDC pa icipan s.
These da a we e collec ed ac oss 35 di e en scanning coho s. Fo
each pa icipan , bo h ana omical and unc ional magne ic esonance
imaging (MRI) da a we e used. Acquisi ion pa ame e s o each scan-
ning si e a e ound a h p:// con 1000.p ojec s.ni c.o g/indi/abide. Addi-
ionally, we ex ac ed se e al composi e sco es om he Au ism Dia-
gnos ic Obse a ion Schedule-Gene ic, Au ism Diagnos ic In e iew-
Re ised, Vineland Adap i e Beha io Scales, Social Responsi eness
Scale, Social Communica ion Ques ionnai e, and aw sco e o he
Au ism Quo ien , and he e bal, pe o mance, and Full Scale IQ
sco es o add ess cogni i e pe o mance and diso de se e i y.
Da a quali y-assu ance
We disca ded subjec s wi h a scanning du a ion sho e han 5 minu es
a e sc ubbing, lacking ull b ain co e age, and an a e age ame-
wise displacemen g ea e han 0.3 mm [69]. Subjec s om he KUL
sample 3 and NYU sample 2 we e also omi ed because hey only
con ained ASD subjec s and he e o e hose coho s did no p o ide
Me hods 77
any TDC. Consequen ly, he numbe o inally included subjec s was
1541 (884 TDC, 657 ASD), co esponding o 33 di e en scanning
coho s ha we e u he me ged in o 24 ins i u ions ollowing he
guidelines p o ided in he ABIDE websi e. The desc ip i e s a is -
ics pe ins i u ion (numbe o subjec s, ASD cases, mean age, sex
dis ibu ion) can be ound in Table 4.1.
Ins i u ion name Numbe o subjec s
con ibu ed
Age
(mean ±sd)
Sex dis ibu ion
(Female)
Numbe o
ASD cases
BNI (II) 40 39.85 ±15.59 0 19
CALTECH (I) 37 27.42 ±9.76 8 18
CMU (I) 20 25.45 ±5.29 4 8
EMC (II) 27 8.40 ±1.09 4 15
ETH (II) 26 22.91 ±4.57 0 6
GU (II) 76 10.92 ±1.66 28 29
IU (II) 37 24.43 ±7.59 9 17
KKI (I and II) 205 10.29 ±1.28 70 48
LEUVEN (I) 61 18.18 ±4.97 7 26
MAX MUN (I) 44 28.77 ±11.79 7 18
NYU (I and II) 245 13.82 ±6.74 42 115
OHSU (II) 83 10.95 ±2.05 31 33
OLIN (I) 22 17.41 ±3.70 4 12
ONRC (II) 18 22.39 ±3.58 5 7
PITT (I) 43 19.53 ±6.87 7 22
SBL (I) 25 34.08 ±6.41 0 13
SDSU (I and II) 86 13.75 ±2.67 15 42
STANFORD (I) 19 10.23 ±1.48 6 11
TCD (I and II) 73 16.86 ±3.41 0 33
UCD (II) 28 14.98 ±1.81 7 15
UCLA (I and II) 70 12.98 ±2.42 8 36
UM (I) 107 14.60 ±3.26 24 41
USM (I and II) 103 23.36 ±7.67 5 52
YALE (I) 46 12.97 ±2.97 13 21
ALL 1541 16.50 ±8.82 304 657
Table 4.1: Main da a cha ac e is ics o each Ins i u ion pa icipa ing in ou s udy.
BNI = Ba ow Neu ological Ins i u e; CALTECH = Cali o nia Ins i u e o Technology; CMU
= Ca negie Mellon Uni e si y; EMC = E asmus Uni e si y Medical Cen e Ro e dam; ETH =
ETH Z¨u ich; GU = Geo ge own Uni e si y; IU = Indiana Uni e si y; KKI = Kennedy K iege
Ins i u e; LEUVEN = Uni e si y o Leu en; MAX MUN = Ludwig Maximilians Uni e si y
Munich; NYU = NYU Langone Medical Cen e ; OHSU = O egon Heal h and Science Uni e si y;
OLIN = Olin; Ins i u e o Li ing a Ha o d Hospi al; ONRC = Olin Neu opsychia y Resea ch
Cen e , Ins i u e o Li ing a Ha o d Hospi al; PITT = Uni e si y o Pi sbu gh School o
Medicine; SBL = Social B ain Lab, Ne he lands Ins i u e o Neu osciences; SDSU = San
Diego S a e Uni e si y; STANFORD = S an o d Uni e si y; TCD = T ini y Cen e o Heal h
Sciences; UCD = Uni e si y o Cali o nia Da is; UCLA = Uni e si y o Cali o nia Los Angeles;
UM = Uni e si y o Michigan; USM = Uni e si y o U ah School o Medicine; YALE = Yale
Child S udy Cen e . I: ABIDE 1. II: ABIDE 2.
78 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
4.2.2 Func ional Connec i i y Ma ices
A s a e-o - he-a p e-p ocessing pipeline was adop ed using FSL
5.0.9, AFNI 16.0.01 [286] and MATLAB 2020b. We i s applied
slice- ime co ec ion, olume alignmen o he a e age one o co ec
o head mo ion a i ac s, which was ollowed by in ensi y no maliz-
a ion. We nex eg essed ou 24 mo ion pa ame e s, as well as he
a e age ce eb ospinal luid (CSF) and a e age whi e ma e signal. A
band-pass il e was applied be ween 0.01 and 0.08 Hz, and linea and
quad a ic ends we e emo ed. All oxels we e spa ially smoo hed
wi h a 6 mm FWHM.
A e p ocessing he s- MRI images, F eeSu e e sion 5.3.0 was
used o b ain segmen a ion and co ical pa cella ion. A o al o 82 e-
gions we e gene a ed om he Desikan-Killiany a las, wi h 68 co ical
egions (34 in each hemisphe e) and 14 subco ical egions segmen ed
om F eeSu e (le / igh halamus, cauda e, pu amen, pallidum,
hippocampus, amygdala, and accumbens). Fo each pa icipan , he
pa cella ion was p ojec ed o he indi idual unc ional da a and he
mean unc ional ime se ies o each egion was ob ained. Finally,
one connec i i y ma ix o each pa icipan was buil by Fishe ’s
Z- ans o ma ion o he Pea son co ela ion coe icien s be ween he
egion pai s o he ime se ies.
4.2.3 Da a Ha moniza ion
We e i ied he p esence o he e ogenei y ela ed o scanning ins i u-
ions in ou connec i i y ma ices by applying o each link a K uskal-
Wallis es , o es ing median loca ion di e ences, and a La ene’s
es o di e ences in a iances. The o me yielded all he exis ing
links signi ican ly di e en ac oss ins i u ions a e FDR co ec ion
(3321 links in o al), whe eas he la e ga e 1236 su i ing links.
To ha monize ou mul i-ins i u ion unc ional connec i i y da a,
and be o e pe o ming sub yping, we used an in-house implemen a-
ion o Comba h ps://pypi.o g/p ojec /pycomba , adjus ing hese mul i-
ins i u ion ba ch e ec s by linea mixed modeling and he use o
empi ical Bayes me hods [282]. We also included in his model he
Me hods 79
diagnosis label (TDC o ASD) as a biological a iable o in e es , en-
su ing ha g oup-le el connec i i y di e ences we e p ese ed a e
ha moniza ion.
The scanning ins i u ion he e ogenei y disappea ed a e Comba
ha moniza ion, while e aining he be ween-g oup a iabili y o ou
da a.
4.2.4 ASD Sub yping ia Consensus Clus e ing
We eg essed ou om he ASD ha monized connec i i y ma ices
he e ec s o age, sex and head mo ion. Then, hese ma ices we e
used o de ine, o each b ain egion i, a ma ix o Euclidean dis ances
be ween uand ASD subjec s, i.e
Di≡di
u =
u
u
M
X
j=1
(yu
ij −y
ij)2,(4.1)
whe e M e lec s he numbe o b ain egions, and yij he whole-
b ain connec i i y pa e n o a gi en egion i(a ec o o dimension
equal o he numbe o egions, whe e each componen is de ined as
he amoun o connec i i y be ween he gi en egion and any o he
in he pa cella ion). Then, each dis ance ma ix Diwas pa i ioned
in o kg oups o subjec s using a k-medoids clus e ing me hod [287],
and he esul ing clus e ing in o ma ion encoded in o an adjacency
ma ix, whose en ies a e 1 i a pai o subjec s belongs o he same
clus e and ze o o he wise. Subsequen ly, a N×Nconsensus ma -
ix Cwas e alua ed by a e aging his in o ma ion ac oss he nodes.
Hence, he en ies o Cu indica e he numbe o pa i ions in which
subjec s uand a e assigned o he same g oup, di ided by he
numbe o pa i ions. E en ually, he consensus ma ix is a e aged
o e he k ange in he in e al (2-20), so ha in o ma ion abou he
unde lying s uc u e a di e en esolu ions is combined in he inal
consensus ma ix, o mo e de ails see [278]. The consensus ma ix
Cwas u he used o de ine a Newman and Gi an-like modula i y
ma ix [288]:

80 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
B=C−P, (4.2)
whe e Pis he expec ed co-assignmen ma ix, uni o m as a con-
sequence o he null ensemble s a egy ob ained by epea ing he pe -
mu a ion o labels 1000 imes. Such a modula i y ma ix Bencodes
all he in o ma ion abou he in e ac ion be ween subjec s a di e -
en le els. As a esul , one could now de ine any dis ance quan i y
applied o his ma ix o assessing clus e ing. Ins ead, we di ec ly
ed his Bma ix in o a gene alized Lou ain me hod o communi y
de ec ion (h ps://gi hub.com/GenLou ain/GenLou ain), yielding an op imal
ou pu pa i ion ha maximizes he ne wo k modula i y.
The s abili y o each sub ype and he 95% CIs o he es ima ed
maximum modula i y we e assessed by boo s apping [289].
4.2.5 S a is ical Di e ences in B ain Mo phology
and Beha io Be ween ASD Sub ypes
We applied mul iple linea eg ession o assess s a is ical di e ences
be ween ASD sub ypes in egion-wise olume and hickness om
F eeSu e , while con olling o age, sex, and o al in ac anial olume
and a one-way analysis o a iance o di e ences in beha io . Mul-
iple es ing was co ec ed by con olling he alse disco e y a e
(FDR).
4.2.6 Func ional connec i i y al e a ions o ASD
sub ypes wi h espec o TDC
To assess he unc ional connec i i y al e a ions be ween each ASD
sub ype and he TDC g oup, we pe o med Mul i a ia e Dis ance
Ma ix Reg ession (MDMR) [290, 291]. Speci ically, MDMR eg essed
each dis ance ma ix pe egion gi en by Eq. 4.1 on o a design ma ix
X o med by a se o mp edic o s, yielding a pseudo-F s a is ic Fi
x
ha eads:
Me hods 81
Fi
x= (HxGi)/(mx)
[(I−H)Gi]/(N−m),(4.3)
whe e indica es he ace ope a o , N he numbe o obse a ions,mx
he deg ees o eedom o p edic o x,Hx he isola ed e ec o p e-
dic o x om he usual ma ix H=X(XTX)−1XT, and Gi he
so-called Gowe ma ix buil om he dis ance ma ix Di[292]. In
ou case, he p edic o s consis ed o he ASD s TDC g oup ac o as
he a iable o in e es , and sex, age, mean amewise displacemen
and FIQ as co a ia es. Like he F-es ima o in a s anda d ANOVA
analysis, Eq. 4.3 assesses he a iance explained by a p edic o a i-
able wi h espec o he unexplained a iance. Finally, o es ima e
how much a iabili y can be a ibu ed o each p edic o , a pseudo-
R2e ec size can be compu ed by di iding he nume a o wi hou he
deg ees o eedom in Eq. 4.3 by he o al sum o squa ed pai wise
dis ances in he Gowe ma ix
R2
x= (HxG)
(G),(4.4)
Simila o s anda d linea models, his e ec size quan i ies he
p opo ion o he o al sum o squa es ha can be explained by he
p edic o s.
4.2.7 T ansc ip omics
To build b ain ansc ip ion maps, we ook ad an age o he publicly
a ailable da a in he AHBA [19]. The da ase consis ed o MRI im-
ages, and a o al o 58,692 mic oa ay-based ansc ip ion p o iles o
abou 20,945 genes sampled om 3,702 di e en egions ac oss he
b ains o six humans. To pool all he ansc ip ion da a in o a single
b ain empla e, we ollowed a simila p ocedu e o ha employed
elsewhe e [25], and p e iously applied in chap e 3, which includes:
(1) P obe e-anno a ion using a e-anno a o oolki [182]; (2) Re-
mo al o p obes whose sampling p opo ion in any o he six b ains
did no exceed he 70%; (3) Unique p obe o gene assignmen using
82 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
he maximum di e en ial s abili y (DS) c i e ion [26]; (4) Remo al o
he in e -subjec di e ences by pooling oge he he Z-sco es o he
ansc ip ion alues o each gene and b ain; (5) Compu a ion o a
single ansc ip ion alue o each egion in he Desikan-Killiany a las
[83] by calcula ing he median o all he alues belonging o he gi en
egion; (6) Gene il e ing conside ing only b ain-speci ic genes ela i e
o o he issues using he Human P o ein A las [11, 293] (h ps://www.
p o eina las.o g). The p ecise lis o b ain-speci ic genes was ha ex ac-
ed on 27 h Sep embe 2021, and i is a ailable on he h ps://gi hub.
com/compneu obilbao/asd-sub yping-en ichmen / ee/main/da a/b ain-speci ic genes.
x
Some me hodological conside a ions need o be cla i ied. Fi s ,
he choice SP > 70%, used in [40] (and chap e 3) was aken ollow-
ing he a ionale o choosing he h eshold as high as possible, bu
wi hou being oo es ic i e, and keeping as many genes as possible
o he inal s udy. Second, he choice o he Desikan-Killiany pa -
i ion was aken as Dono ’s b ains can be di ec ly segmen ed using
F eeSu e , which inco po a es by de aul he Desikan-Killiany pa -
i ion. To use a di e en pa i ion, i is necessa y o ans o m in o
ha empla e he dono ’s b ain, whose images we e acqui ed ex- i o
and as such, i is e y likely o con ain mo e misalignmen s in he
b ain samples. Finally, es ic ing ou analyses o genes ha a e el-
e an in he s udy o he b ain was done o p eselec a subse o all
genes, as AHBA includes genes exp essed in any body issue. A e
all he a o emen ioned conside a ions, he esul ing o al numbe o
b ain-speci ic genes was 1882.
4.2.8 The use o spa ial au o eg essi e models o
he associa ion o ansc ip omics wi h sub-
ypes
The associa ion be ween b ain maps migh be in la ed due o he
exis ence o spa ial au oco ela ions. To con ol o his excess o
biased co ela ions, we in oduced a spa ially lagged dependen a i-
able o ou eg ession models, i.e. we conside ed a model o he o m
Me hods 83
y=ρWy+Xβ+ , whe e y is he b ain pa e n o ansc ip omics o
a pa icula gene, X he inpu ma ix ha con ains he in e cep and
he pseudo-R2maps o a pa icula sub ype, Wa weigh ma ix ha
de e mines he spa ial au oco ela ion s uc u e, and he no mally
dis ibu ed model’s noise. Bo h βand ρa e he es ima es, wi h he
o me quan i ying he deg ee o e ec o Xin y, and he la e he
amoun o spa ial au oco ela ion. Such a model was implemen ed
using he unc ion ML Lag in he pysal package (h ps://pysal.o g/). To
de ine he weigh ma ix W, a dis ance-based app oach was adop ed,
such ha spa ial au oco ela ions we e la gely d i en by nea es b ain
egions. The co ico-co ical dis ance was calcula ed by using hei
geodesic dis ance implemen ed in B ainSmash package [185], while
o de e mine he sobco ico-co ical dis ance, and simila o [29], he
Euclidean dis ance was calcula ed [29, 185, 294].
As a esul , o each gene we ob ained one -s a is ic and one p-
alue, which allowed us o assess he associa ion wi h he pseudo-R2
maps while accoun ing o possible spa ial au oco ela ions. Among
he signi ican ly associa ed genes, we iden i ied as ele an hose genes
included in he SFARI da abase.
4.2.9 Gene Se En ichmen Analysis and P o ein
In e ac ion Analysis
We only conside ed o he analyses such genes wi h FDR-co ec ed p
(pFDR) alue < .05 in each sub ype. A e ha , we pe o med a gene
se en ichmen analysis using WebGes al [295] (h p://www.webges al .o g/),
in oducing as he inpu he lis o he co ec ed genes and he -
s a is ic om he associa ion analysis. We compu ed he gene se
en ichmen analysis o gene on ology (GO) biological p ocess [31]
and Reac ome pa hways [221] and only conside ed en iched ca ego ies
pFDR alue < .05. We u he applied an ensemble-based en ichmen
analysis, simila o he one de eloped in [296], o e alua e whe he
signi ican en ichmen anno a ions we e a ec ed by in la ion o alse
posi i e bias [296]. Fi s , we gene a ed 10,000 su oga e b ain maps
wi h he same spa ial au oco ela ion as he o iginal pseudo-R2maps
90 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
Nex , we assessed he di e ences in connec i i y pa e ns be ween
each ASD sub ype and he TDC g oup, measu ed by egionwise no -
malized pseudo-R2b ain maps, esul ing om mul i a ia e dis ance
ma ix eg ession (Figu e 4.4). The spa ial simila i y be ween hese
maps was e y low ( 80 = 0.09, pe mu a ion-based p= .67, a e
using 5000 su oga es ha p ese ed spa ial au oco ela ion), indic-
a ing ha each sub ype exhibi ed a dis inc neu obiological p o ile o
b ainwide connec i i y, as expec ed since he sub ypes we e ob ained
by clus e ing he unc ional connec i i y p o iles. Speci ically, o sub-
ype 1, highe di e ences as compa ed wi h TDC we e ound in he
supe io empo al gy us, pos e io cingula e co ex, and he insula,
co e ing he unc ional ne wo ks o de aul mode and salience. Fo
sub ype 2, highe di e ences exis ed in he halamus, simila o p e-
ious wo k [301], pu amen, and p ecen al gy us. Thus, al e a ions
a ec ing he de aul mode ne wo k we e common o bo h sub ypes,
bu one (sub ype 1) also showed speci ic dis up ions in ol ing he
salience ne wo k and he o he (sub ype 2) in he soma omo o ne -
wo k.
Fo he biological cha ac e iza ion o each sub ype, we se ou o
iden i y which genes had an exp ession ac oss b ain egions signi ic-
an ly associa ed (pFDR <.05) wi h he di e ences in connec i i y
measu ed by he no malized R2b ain maps (Figu e 4.4, his og ams),
whe eby la ge R2 alues co espond o la ge unc ional connec i i y
al e a ions. Fo sub ype 1, a o al o 195 nega i e-associa ed (NEG)
genes and 364 posi i e-associa ed (POS) genes exis ed. Signi ican
NEG genes, also p esen in he SFARI gene human da abase wi h
a ele ance sco e o 1, we e GFAP, CHD7, SKI, SHANK3, ANK3,
and CACNA1E, while POS genes we e ASXL3, MAP1A, STXBP1,
DPYSL2, KNCB1, SCN8A, RIMS1, and CDKL5. Simila ly, o sub-
ype 2, we ound 142 NEG genes, o which GRIA2, RFX3, SHANK2,
GRIN2B, DLG4, LRRC4C, ARX, and GABRB3 we e also p esen in
he SFARI lis , and 180 POS genes, including MAGEL2 and IQSEC2.
We nex applied gene en ichmen o he lis o signi ican genes wi hin
each sub ype, inding no signi ican en ichmen o sub ype 1, he ype

Resul s 91
Figu e 4.4: Associa ion be ween ansc ip omics and connec i i y pa e ns o each
au ism spec um diso de sub ype. Fo sub ypes 1 and 2, we calcula ed he pseudo-
R2map, conside ing he di e ences in he connec i i y pa e n ha each sub ype has om
ypically de eloping con ol pa icipan s. (Righ ) B ain maps o no malized pseudo-R2. (Le )
His og ams o associa ion alues be ween pseudo-R2and gene ansc ip ion ac i i y (di e en
alues co espond o associa ion wi h di e en genes). This p ocedu e was epea ed using he
pseudo-R2map o each sub ype. The ail o he nega i e genes ( alse disco e y a e–co ec ed
p<.05 and -s a is ic <0) is ma ked wi h a blue ec angle and he ail o he posi i e genes
( alse disco e y a e–co ec ed p<.05 and -s a is ic >0) wi h a ed one o bo h sub ypes.
Signi icance limi s ( ) a e also shown. Fo each dis ibu ion ail, we also show he ele an genes
p esen in he SFARI au ism spec um diso de genes wi h a sco e = 1.
wi h b ain hypoconnec i i y. Howe e , o sub ype 2, he en ichmen
o he NEG genes included GO biological p ocesses and Reac ome
pa hways ela ed o glu ama e signaling (a ec ing bo h AMPA and
NMDA ecep o s) and synapse o ganiza ion in ela ion o he E/I
imbalance occu ing du ing he de elopmen o b ain ci cui s (Fig-
u e 4.5, panel A). We also assessed which NEG genes pa icipa ed
in each biological p ocess and pa hway (Figu e 4.5, panel B), inding
ha genes DLG4, GRIN2B, GRIA2, and SHANK2 we e pa icipa -
ing in mos o hem; and, he gene DLG4 plays a ole in all o hem.
Addi ionally, he DLG4 gene was he one wi h he highes deg ee in
he p o ein in e ac ion ne wo k.
We ound a signi ican en ichmen wi h biological p ocesses e-
la ed o E/I imbalance o sub ype 2 bu no o sub ype 1. To es
92 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
Figu e 4.5: Exci a ion/inhibi ion imbalance en ichmen o only one class o pa -
icipan s wi h au ism spec um diso de (sub ype 2).A: GSEA cha ac e iza ion o he
alse disco e y a e–signi ican genes in sub ype 2, including he GO biological p ocesses (da k
g ay) and Reac ome pa hways (ligh g ay) en ichmen s. We u he es ed whe he he en ich-
men indings we e a ec ed by hei epo ing a e in he li e a u e, and o all cases epo ed
he e, we ob ained pFDR <.05. B: Pa icipa ion coun o each gene in he p ocesses shown in
(A) anging om 4 o 10 (co esponding o a pa icipa ion in all p ocesses ha only occu ed
o DLG4). C: P o ein-p o ein in e ac ion physical ne wo k om he lis o alse disco e y
a e–signi ican genes. Fo ease o isualiza ion, only subne wo ks wi h a minimum o 10 genes
a e depic ed. D: Node deg ee o he genes pa icipa ing in he ne wo k shown in (C). DLG4
is he gene wi h he highes deg ee. (B–D) Ba s co esponding o genes wi h SFARI sco e = 1
a e colo ed in ed; SFARI sco e = 1S in da k ed; SFARI sco e = 2 in o ange; and SFARI sco e
= 2S in da k o ange, and he same colo code was used in (C) and (D) o ne wo k nodes.
whe he hese indings sugges ed ha he unc ional connec i i y in
sub ype 1 was di e en om ha in p e ious s udies o ASD, we
calcula ed he simila i y o he connec i i y p o iles o ou 2 sub-
ypes wi h ypical connec i i y al e a ions in ASD, ep esen ed by
b ain maps in [302] and calcula ed om 4 di e en ASD da abases
o es ing unc ional MRI da a. We calcula ed o each sub ype he
a e age spa ial simila i y ac oss he 4 exis ing b ain maps o con-
nec i i y al e a ions in [302]. Fo sub ype 1 he a e age simila i y
was no signi ican ( 80 = 0.19, p =.45), bu o sub ype 2 i was sig-
ni ican ( 80 = 0.46, p =.02), indica ing ha sub ype 2 mo e closely
esembled he ypical connec i i y al e a ions epo ed in ASD, which
Resul s 93
in ac is he sub ype o which we ound signi ican en ichmen o-
wa d E/I imbalance.
Robus ness o ansc ip omic-connec i i y associa ion es-
ul s
We also compa ed he esul s ob ained om he gene alized Lou ain
algo i hm o hose ound by mul i esolu ion clus e ing. As a measu e
o simila i y be ween he 2 solu ions, Dice index alues o he solu-
ions we e 0.99 and 0.89, espec i ely, which indica ed a high le el o
ep oducibili y o he gene exp ession associa ion wi h b ain al e a-
ions be ween he 2 clus e ing s a egies.
Addi ionally, we s udied he e ec s o conside ing a di e en b ain
pa i ion on he esul s o he ansc ip ion-connec i i y associa ion.
Using he unc ionally de ined Schae e [82] b ain pa i ion wi h 100
di e en egions, he associa ion esul s ob ained om he Desikan-
Killiany a las as compa ed o hose om he Schae e pa i ion had
e y low simila i y o sub ype 1 ( 1880 =−0.11, p < .001), and
sligh ly highe esul s we e ound o sub ype 2 ( 1880 = 0.40, p <
.001). By adding he same subco ical egions o he Schae e pa -
i ion, he gene associa ion became e y simila o he 2 b ain pa -
i ions and o he 2 sub ypes (sub ype 1, 1880 = 0.87, p < .001;
sub ype 2, 1880 = 0.92, p < .001), sugges ing a s ong con ibu ion
o he subco ical al e a ions o he obus ness o ou associa ion
esul s. These esul s we e ob ained by using le -hemisphe e an-
sc ip ion si es 2, bu he esul s we e also p ese ed when we epea ed
he analysis o he 2 b ain hemisphe es (Figu e 4.6).
2We ocused on he le hemisphe e, as all dono s p o ided sampling si es o
genes in his hemisphe e, and only 2 o he 6 dono s om he AHBA da ase we e
sampled in bo h le and igh hemisphe es.
94 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
Figu e 4.6: Replicabili y o he esul s including samples om he whole b ain.
Simila o Figu e 4.5, bu now including gene ic in o ma ion om sampling si es in bo h b ain
hemisphe es. These new esul s show eplicabili y o he epo ed en ichmen , al hough mo e
genes a e now implica ed in addi ion o an inc eased complexi y o he p o ein-p o ein in e ac-
ion ne wo k.
Finally, i is impo an o no e ha no signi ican en ichmen was
ound o sub ypes lowe in he dend og am le el co esponding o
he 2 sub ypes desc ibed abo e (Figu e 4.2). The signi ican en ich-
men did no exis a e epea ing he same analysis using he en-
i e ASD g oup, indica ing he need o sub yping i s in he en i e
Discussion 95
popula ion o e eal ou indings. To p o e ha ou gene en ichmen
indings we e speci ic o he ASD condi ion we also epea ed he same
p ocedu e bu only using he TDC popula ion. To do ha , we i s
di ided he en i e TDC coho in o wo subg oups, hal -sized and an-
domly chosen, so ha hey we e well-ma ched o age, sex, mo emen ,
FIQ, and o e all connec i i y (mean ac oss posi i e en ies in he
connec i i y ma ices). Nex , we applied in one subg oup he same
sub yping p ocedu e (including he same p e ious denoising s ep o
a oid clus e s d i en by e ec s o no in e es ), ob aining again wo
sub ypes, one ep esen ing hype -connec i i y and he o he hypo-
connec i i y. We hen used he o he TDC subg oup o calcula e
he pseudo-R2s a is ical maps, which we e subsequen ly associa ed
wi h he ansc ip omics da a. As a esul , no gene su i ed FDR
co ec ion in any sub ype, hus indica ing ha he E/I imbalance
ound in he hype connec ed au is ic sub ype is speci ic o he au -
is ic condi ion. Likewise, al hough he sub yping pe o med in bo h
ASD and TDC g oups esul ed in solu ions wi h simila o e all con-
nec i i y sepa a ion, esul ing in 2 se s o hypo- and hype connec ed
b ains, a mul i a ia e dis ance ma ix eg ession analysis applied o
he unc ional co ela ion pa e ns showed ha he hype connec ed
sub ype ound in ASD was s a is ically di e en om ha in TDC
(p<.001). The hypoconnec ed sub ypes in ASD and TDC we e
also di e en om each o he (p<.001). This migh explain why no
simila indings in he en ichmen we e ound o he hype connec ed
TDC subg oup. In summa ion, he signi ican associa ion be ween
E/I imbalance and al e ed unc ional connec i i y was obse ed when
sub yping in ASD, and only in he ASD g oup cha ac e ized by o e -
all hype connec i i y, demons a ing he speci ici y o he epo ed
en ichmen .
4.4 Discussion
Two signi ican sub ypes esul om unc ional connec i i y–based
sub yping in a coho o 657 indi iduals wi h ASD. The wo a e in-
dis inguishable by beha io al sco es, and also by mo phome ic com-

96 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
pa isons based on s uc u al neu oimaging, in ag eemen wi h ecen
esul s [275]. Compa ed wi h he TDC g oup, he i s sub ype is
cha ac e ized by hypoconnec i i y, wi h majo implica ions in he
supe io empo al gy us, pos e io cingula e co ex, and insula, show-
ing connec i i y al e a ions in he de aul mode and salience ne wo ks
wi h no signi ican gene en ichmen a e co ec ing o mul iple com-
pa isons. The second sub ype, ep esen ing 43% o pa icipan s wi h
au ism, is cha ac e ized by hype connec i i y, wi h majo implica-
ions in he halamus, pu amen, and p ecen al gy us and showing
ne wo k al e a ions in soma omo o and de aul mode ne wo ks. In
a ecen analysis linking genomics and es ing unc ional connec i -
i y in 32,726 indi iduals wi h psychia ic condi ions, signi ican ASD
con ibu ions we e shown in he halamic and soma omo o ne wo ks
[301], consis en wi h ou esul s o sub ype 2. Only sub ype 2 had
a signi ican gene en ichmen owa d glu ama e signaling (a ec ing
bo h AMPA and NMDA ecep o s), consis en wi h he E/I imbal-
ance ha occu s du ing b ain de elopmen and one o he mos ac-
cep ed hypo heses in he pa hophysiology o au ism [303]. Indeed, i
is hough ha in he de elopmen o ASD, he e is an inc ease in
he a io be ween exci a ion and inhibi ion, leading o hype exci ab-
ili y o co ical ci cui s [249]. I is also possible ha di e en ial E/I
al e a ion o selec i e b ain ci cui s migh esul in an unal e ed E/I
a io a he ne wo k le el [248]. Ou wo k maps pa e ns o unc-
ional connec i i y al e a ions wi h genes ha a e in ol ed in E/I
balance. While i is ue ha pe u ba ion in hese genes in animal
models s ongly a ec s E/I imbalance in b ain ne wo ks [304], he
pa icipan da a ha we analyzed in his s udy do no di ec ly ad-
d ess he E/I imbalance, and his is a limi a ion o ou me hodology.
I is also impo an o emphasize ha he E/I en ichmen ound in
ou s udy is speci ic o he ASD condi ion and, as such, does no
occu in he TDC g oup. Mo eo e , he connec i i y p o ile in he
en i e au is ic popula ion, i.e., i no sub yping is pe o med, does no
ha e signi ican en ichmen , indica ing he need o sub yping i s o
ind he connec ion wi h E/I imbalance in one sub ype o indi iduals
wi h ASD.
Discussion 97
Ou sub yping app oach was based on pa e ns o unc ional con-
nec i i y al e a ions. The e a e 3 majo easons suppo ing ou choice
no o use s uc u al ea u es o ou sub yping analysis. Fi s , i
would equi e a di e en clus e ing app oach o he one adop ed he e,
which is based on he consensus o connec i i y pa e ns. Second,
and based on ecen da a-d i en esul s om an in e na ional au ism
imaging bioma ke challenge [275] wi h mo e han 146 ins i u ions
submi ing p edic ion algo i hms, he 10 bes -pe o ming algo i hms
(wi h ASD p edic ion accu acies ha ing a ea unde he cu e >0.80)
showed a dominan con ibu ion o he unc ional modali y, wi h
a much highe disc imina i e powe han he s uc u al MRI da a.
Thi d, ou main goal was o s udy he o igin o unc ional connec i -
i y–based he e ogenei y in au ism, and s uc u al ea u es ( ep esen -
ing di e en b ain aspec s) gi e ise o a di e en kind o he e ogen-
ei y. As a esul , he p ope combina ion o hese 2 di e se sou ces
o he e ogenei y would equi e a mul imodal app oach di e en om
he one de eloped he e.
Ou app oach is unique in se e al ways. Fi s , ou s udy is based
on a la ge coho o indi iduals wi h ASD (N = 657) om he ABIDE
ini ia i e, all o hem ha ing passed he igo ous c i e ia o mo ion
emo al, and i combines ana omical and unc ional neu oimaging
da a om 24 di e en ins i u ions. Second, we used Comba , a igo -
ous da a ha moniza ion me hod o elimina e he a iabili y be ween
MRI scans ac oss he 24 ins i u ions, one o he la ges sou ces o
a iabili y when combining imaging da a om mul iple ins i u ions
[305]. Thi d, ou analysis o b ain connec i i y was ca ied ou on a
la ge scale, in which each b ain egion is ep esen ed by i s connec i -
i y pa e n ac oss he en i e b ain. The e o e, we did no conside a
p io i any b ain egion as mo e dominan o ele an han he o h-
e s. Fou h, we made use o a consensus clus e ing app oach ha we
de eloped [6, 279], and ha has been success ully es ed by o he s
[306], o g oup pa icipan s in he same sub ype i he connec i i y
p o iles a e simila ac oss all he analyzed egions. Finally, we made
use o he AHBA o desc ibe he neu ogene ic p o iles o each sub-
ype, which has been used be o e o mo phome ic in o ma ion in
98 Chap e 4. The Neu ogene ics o FC Al e a ions in Au ism
ASD [225] bu ne e o cha ac e izing sub ypes based on unc ional
connec i i y pa e ns o his condi ion.
Due o he he e ogenei y and di e si y epo ed in ASD gene -
ics, he use o AHBA may shed new ligh , because i p o ides in-
o ma ion on he ansc ip ome ac oss he b ain in unp eceden ed
de ail, accoun ing o 3702 sampling si es wi h ansc ip ion in o m-
a ion on 20,500 genes as a speci ic signa u e o each ana omical e-
gion. Mo eo e , he use o AHBA is complemen a y o o he ech-
niques, such as GWAS [226], ha simul aneously add ess geno ype-
pheno ype associa ions om hund eds o housands o millions o
gene ic a ian s in a da a-d i en manne . Indeed, genome-wide asso-
cia ion s udies ha e p e iously been used o ASD sub yping [307, 16]
using beha io al sco es as ai s and, he e o e, he sub ypes ob ained
we e mo e closely ela ed o symp om se e i y and no o unc ional
connec i i y.
Ou en ichmen esul s o sub ype 2 show ha DLG4, also known
as PSD95, is a gene wi h majo implica ions in he p o ein in e -
ac ion ne wo k o sub ype 2. DLG4 media es NMDA and AMPA
ecep o clus e ing and unc ion; i a ec s glu ama e gic ansmis-
sion and has been shown o ha e an abe an unc ion in ASD [308,
309, 310, 311, 312]. DLG4 also in luences he size and densi y o
dend i ic spines du ing b ain de elopmen , ha ing s ong e ec s on
synap ic connec i i y and ac i i y, e.g., educed DLG4 ac i i y leads
o inc eased dend i ic spine numbe s [313].
Some limi a ions should be no ed. Fi s , ou ansc ip omic ana-
lysis was based on AHBA, which is de i ed om heal hy, and no om
ASD, b ain issues. The e o e, he ela ions s udied he e be ween
ASD-dependen connec i i y pa e ns and heal hy ansc ip omics
highligh la ge-scale o ganiza ion aspec s o he connec i i y al e -
a ions o gene exp ession. Fu u e s udies should con i m ou indings
using gene exp ession da a om a pa hological coho , which is no
cu en ly a ailable. Second, he numbe o dono s om he AHBA
da a is e y limi ed (n = 6) and he sampling si es a ailable do no
co e he ull b ain. Thi d, ou sub yping me hod ound 2 sub ypes
o ASD pa icipan s who we e hypoconnec ed and hype connec ed
Da a and code a ailabili y 99
a he ne wo k le el. The same classes o sub ypes we e ound by
sub yping he TDC g oup. Howe e , when compa ing i s he hypo-
connec i i y sub ypes be ween he ASD and TDC g oups and hen
he hype connec i i y sub ypes, connec i i y pa e ns we e signi ic-
an ly di e en in bo h cases, which jus i ies he signi ican en ichmen
ound o he hype connec i i y ASD sub ype, bu no o TDC. Fi-
nally, ou main neu ogene ic inding in one ASD sub ype, in ol ing
genes la gely a ec ing he E/I imbalance, is based only on he s a is -
ical associa ion be ween ansc ip ome ac i i y and pa e ns o unc-
ional connec i i y al e a ions. Fu u e s udies should explici ly es
he causal link be ween E/I imbalance and unc ional connec i i y
al e a ions in ASD.
In summa y, ou no el app oach, which includes da a ha moniz-
a ion, mul i a ia e dis ancing in la ge-scale unc ional connec i i y
pa e ns, and ansc ip ome b ain maps, e eals s ong en ichmen
o glu ama e signaling (a ec ing bo h AMPA and NMDA ecep o s)
and synapse o ganiza ion in one subg oup o pa icipan s wi h ASD,
ein o cing he hypo hesis o an E/I imbalance occu ing du ing b ain
de elopmen o indi iduals wi h ASD.
4.5 Da a and code a ailabili y
The da a employed in his s udy belong o he ABIDE-I and ABIDE-
II eposi o ies. Thei IDs can be ound in h ps://gi hub.com/compneu obilbao/
asd-sub yping-en ichmen , as well as he codes used o he analyses.
The ini ial mul icen e ABIDE da ase consis ed o 35 di e en scan-
ning coho s and wo g oups o subjec s, TDC (N=1130) and ASD
(N=1026). A e ollowing ou da a quali y-assu ance, we ob ained
connec i i y ma ices o 884 TDC and 657 ASD, belonging o 24 ins i-
u ions, which we e ul ima ely used o he sub yping analysis. These
ma ices a e a ailable a h ps://doi.o g/10.6084/m9. igsha e.21901821.
106 Bibliog aphy
[8] M. Ga ica e al. High-o de in e dependencies in he aging
b ain. B ain Connec i i y, 11(9):734–744, 2021.
[9] M. Ga ica e al. High-o de unc ional edundancy in ageing
explained ia al e a ions in he connec ome in a whole-b ain
model. PLOS Compu a ional Biology, 18(9):e1010431, Sep em-
be 2022.
[10] A. Jimenez-Ma in e al. B ain connec i i y and cogni i e
unc ioning in indi iduals six mon hs a e mul io gan ailu e.
Neu oImage. Clinical, 25:102137, 2020.
[11] I. Diez and J. Sepulc e. Un eiling he neu oimaging-gene ic in-
e sec ions in he human b ain. Cu en Opinion in Neu ology,
Publish Ahead o P in , 5 2021. [Online; accessed 2022-02-02].
[12] L. Shen e al. Whole genome associa ion s udy o b ain-wide
imaging pheno ypes o iden i ying quan i a i e ai loci in
MCI and AD: A s udy o he ADNI coho . Neu oImage,
53(3):1051–1063, No embe 2010.
[13] The Alzheime ’s Disease Neu oimaging Ini ia i e e al. Com-
mon gene ic a ian s in luence human subco ical b ain s uc-
u es. Na u e, 520(7546):224–229, Ap il 2015.
[14] D. P. Hiba e al. No el gene ic loci associa ed wi h hippo-
campal olume. Na u e Communica ions, 8(1):13624, Janua y
2017.
[15] L. T. Ellio e al. Genome-wide associa ion s udies o b ain
imaging pheno ypes in UK Biobank. Na u e, 562(7726):210–
216, Oc obe 2018.
[16] A. Yousa e al. Quan i a i e genome-wide associa ion s udy
o six pheno ypic subdomains iden i ies no el genome-wide sig-
ni ican a ian s in au ism spec um diso de . T ansla ional
Psychia y, 10(1):215, 12 2020.

Bibliog aphy 107
[17] B. Zhao e al. T ansc ip ome-wide associa ion analysis o b ain
s uc u es yields insigh s in o pleio opy wi h complex neu o-
psychia ic ai s. Na u e Communica ions, 12(1):2878, May
2021.
[18] X. Bledsoe and E. R. Gamazon. A T ansc ip omic A las o
he Human B ain Re eals Gene ically De e mined Aspec s o
Neu opsychia ic Heal h. medRxi : The P ep in Se e o
Heal h Sciences, pp. 2023.03.10.23287072, June 2023.
[19] M. J. Haw ylycz e al. An ana omically comp ehensi e a las o
he adul human b ain ansc ip ome. Na u e, 489(7416):391–
399, 2012.
[20] J. M. Keil e al. B ain T ansc ip ome Da abases: A Use ’s
Guide. The Jou nal o Neu oscience: The O icial Jou nal o
he Socie y o Neu oscience, 38(10):2399–2412, Ma ch 2018.
[21] H. J. Kang e al. Spa io- empo al ansc ip ome o he human
b ain. Na u e, 478(7370):483–489, Oc obe 2011.
[22] C. Colan uoni e al. Tempo al dynamics and gene ic con-
ol o ansc ip ion in he human p e on al co ex. Na u e,
478(7370):519–523, Oc obe 2011.
[23] J. A. Mille e al. T ansc ip ional landscape o he p ena al
human b ain. Na u e, 508(7495):199–206, Ap il 2014.
[24] GTEx Conso ium e al. Gene ic e ec s on gene exp ession
ac oss human issues. Na u e, 550(7675):204–213, Oc obe
2017.
[25] A. A na ke iciu e e al. A p ac ical guide o linking b ain-wide
gene exp ession and neu oimaging da a. Neu oImage, 189:353–
367, Ap il 2019.
[26] M. Haw ylycz e al. Canonical gene ic signa u es o he adul
human b ain. Na u e Neu oscience, 18(12):1832–1844, 12 2015.
PMID: 26571460 PMCID: PMC4700510.
108 Bibliog aphy
[27] J. Rase o e al. The neu ogene ics o unc ional connec i i y
al e a ions in Au ism: Insigh s om sub yping in 657 pa ien s.
Biological Psychia y, pp. S0006–3223(23)01230–1, Ap il 2023.
[28] I. Diez and J. Sepulc e. Neu ogene ic p o iles delinea e la ge-
scale connec i i y dynamics o he human b ain. Na u e Com-
munica ions, 9(1):3876, Sep embe 2018.
[29] J. B. Bu e al. Hie a chy o ansc ip omic specializa ion
ac oss human co ex cap u ed by s uc u al neu oimaging o-
pog aphy. Na u e Neu oscience, 21(9):1251–1259, 9 2018.
[30] A. Al mann e al. Analysis o b ain a ophy and local gene
exp ession in gene ic on o empo al demen ia. B ain Commu-
nica ions, 2(2): caa122, July 2020.
[31] M. Ashbu ne e al. Gene on ology: ool o he uni ica ion
o biology. he gene on ology conso ium. Na u e Gene ics,
25(1):25–29, 5 2000. PMID: 10802651 PMCID: PMC3037419.
[32] E. Bueichek´u e al. Cen al neu ogene ic signa u es o he isuo-
mo o in eg a ion sys em. P oceedings o he Na ional Academy
o Sciences o he Uni ed S a es o Ame ica, 117(12):6836–6843,
Ma ch 2020.
[33] C. Paquola e al. Shi s in myeloa chi ec u e cha ac e ise ad-
olescen de elopmen o co ical g adien s. eLi e, 8:e50482,
No embe 2019.
[34] J. Sepulc e e al. Neu ogene ic con ibu ions o amyloid be a
and au sp eading in he human co ex. Na u e Medicine,
24(12):1910–1918, Decembe 2018.
[35] M. J. G o he e al. Molecula p ope ies unde lying egional
ulne abili y o Alzheime ’s disease pa hology. B ain: A
Jou nal o Neu ology, 141(9):2755–2771, Sep embe 2018.
Bibliog aphy 109
[36] Y.-Q. Zheng e al. Local ulne abili y and global connec i -
i y join ly shape neu odegene a i e disease p opaga ion. PLoS
biology, 17(11):e3000495, No embe 2019.
[37] S. Basaia e al. Neu ogene ic ai s ou line ulne abili y o co -
ical dis up ion in Pa kinson’s disease. Neu oImage. Clinical,
33:102941, 2022.
[38] H. San ama ´ıa-Ga c´ıa e al. Neu ocogni i e pa e ns ac oss ge-
ne ic le els in beha io al a ian on o empo al demen ia: a
mul iple single cases s udy. BMC neu ology, 22(1):454, Decem-
be 2022.
[39] G. Sha iei e al. Ne wo k s uc u e and ansc ip omic ul-
ne abili y shape a ophy in on o empo al demen ia. B ain,
146(1):321–336, Janua y 2023.
[40] A. Jimenez-Ma in e al. T ansc ip ional signa u es o synap ic
esicle genes de ine myo onic dys ophy ype I neu odegene -
a ion. Neu opa hology and Applied Neu obiology, 47(7):1092–
1108, Decembe 2021.
[41] J. Sun e al. Genes associa ed wi h g ey ma e olume educ-
ion in mul iple scle osis. Jou nal o Neu ology, 269(4):2004–
2015, Ap il 2022.
[42] A. Buchin e al. Mul i-modal cha ac e iza ion and simula ion o
human epilep ic ci cui y. Cell Repo s, 41(13):111873, Decem-
be 2022.
[43] Y. Yin e al. S uc u al and unc ional changes in d ug-na¨ı e
benign childhood epilepsy wi h cen o empo al spikes and hei
associa ed gene exp ession p o iles. Ce eb al Co ex (New Yo k,
N.Y.: 1991), 33(10):5774–5782, May 2023.
[44] A. S. Mandal e al. Gene ic, cellula , and connec omic cha -
ac e iza ion o he b ain egions commonly plagued by glioma.
B ain: A Jou nal o Neu ology, 143(11):3294–3307, Decembe
2020.
110 Bibliog aphy
[45] R. Rome o-Ga cia e al. T ansc ip omic and connec omic co -
ela es o di e en ial spa ial pa e ning among gliomas. B ain:
A Jou nal o Neu ology, 146(3):1200–1211, Ma ch 2023.
[46] Y. Xie e al. B ain mRNA Exp ession Associa ed wi h Co ical
Volume Al e a ions in Au ism Spec um Diso de . Cell Repo s,
32(11):108137, Sep embe 2020.
[47] L.-B. Cui e al. T ansc ip ional le el o in lamma ion ma k-
e s associa es wi h sho - e m b ain s uc u al changes in i s -
episode schizoph enia. BMC medicine, 21(1):250, July 2023.
[48] X. Xu e al. Gene ic mechanisms unde lying g ay ma e
olume changes in pa ien s wi h d ug-nai e i s -episode schizo-
ph enia. Ce eb al Co ex (New Yo k, N.Y.: 1991), 33(5):2328–
2341, Feb ua y 2023.
[49] K. M. Ande son e al. Con e gen molecula , cellula , and co -
ical neu oimaging signa u es o majo dep essi e diso de . P o-
ceedings o he Na ional Academy o Sciences, 117(40):25138–
25149, Oc obe 2020.
[50] J. Li e al. Co ical s uc u al di e ences in majo dep essi e
diso de co ela e wi h cell ype-speci ic ansc ip ional signa-
u es. Na u e Communica ions, 12(1):1647, Ma ch 2021.
[51] P. Manza e al. B ain s uc u al changes in cannabis de-
pendence: associa ion wi h MAGL. Molecula Psychia y,
25(12):3256–3266, Decembe 2020.
[52] I. Diez e al. Ea ly-li e auma endopheno ypes and b ain
ci cui -gene exp ession ela ionships in unc ional neu ological
(con e sion) diso de . Molecula Psychia y, 26(8):3817–3828,
Augus 2021.
[53] A. L. de Munain e al. P e alence o myo onic dys ophy in
guipuzcoa (basque coun y, spain). Neu ology, 43(8):1573–1573,
8 1993.
Bibliog aphy 111
[54] F. J. Gall. Le e om D . F. J. Gall, o Joseph F on Re ze ,
upon he Func ion o he B ain, in Man and Animals. De neue
Teu sche Me ku , 3:311–332, 1798.
[55] J. G. Spu zheim. The Ana omy o he B ain, wi h a Gene al
View o he Ne ous Sys em. Bos on: Ma sh, Capen & Lyon,
1834, 1826.
[56] P. P. B oca. Su le p incipe des localisa ions c´e ´eb ales. Bull.
de la Soci´e ´e d”An h opologie, 2:190–204, 1861.
[57] P. P. B oca. Localisa ions des onc ions c´e ´eb ales. si`ege de la
acul ´e du langage a icul´e. Bull. de la Soci´e ´e d”An h opologie,
4:200–208, 1863.
[58] J. P. Flou ens. Reche ches exp´e imen ales su les p op i´e ´es e
les onc ions du sys `eme ne eux dans les animaux e ´eb ´es.
Pa is: JB Balli`e e, 1842.
[59] A. D. Boes e al. Ne wo k localiza ion o neu ological symp oms
om ocal b ain lesions. B ain: A Jou nal o Neu ology, 138(P
10):3061–3075, 2015.
[60] J. Jou sa e al. Lesion ne wo k mapping o symp om local-
iza ion: ecen de elopmen s and u u e di ec ions. Cu en
Opinion in Neu ology, 35(4):453–459, aug 2022.
[61] C. Ganos e al. A neu al ne wo k o ics: insigh s om causal
b ain lesions and deep b ain s imula ion. B ain, 145(12):4385–
4397, 01 2022.
[62] J. S. Siegel e al. Dis up ions o ne wo k connec i i y p edic
impai men in mul iple beha io al domains a e s oke. P o-
ceedings o he Na ional Academy o Sciences, 113(30), July
2016.
[63] A. Sal alaggio e al. Pos -s oke de ici p edic ion om lesion
and indi ec s uc u al and unc ional disconnec ion. B ain,
143(7):2173–2188, July 2020.

112 Bibliog aphy
[64] A. Jimenez-Ma in e al. Mul imodal and mul idomain lesion
ne wo k mapping enhances p edic ion o senso imo o beha io
in s oke pa ien s. Scien i ic Repo s, 12(1):22400, Decembe
2022.
[65] O. Spo ns e al. The human connec ome: A s uc u al desc ip-
ion o he human b ain. PLoS Compu . Biol, 1:245–251, 2005.
[66] T. J. Sejnowski e al. Pu ing big da a o good use in neu -
oscience. Na u e Neu oscience, 17(11):1440–1441, No embe
2014.
[67] K. K upa and M. Bekiesi´nska-Figa owska. A i ac s in mag-
ne ic esonance imaging. Polish Jou nal o Radiology, 80:93–
106, 2015.
[68] R. A. Pold ack. Handbook o Func ional MRI Da a Analysis.
Camb idge Uni e si y P ess, 2011.
[69] J. D. Powe e al. Spu ious bu sys ema ic co ela ions in
unc ional connec i i y m i ne wo ks a ise om subjec mo ion.
Neu oImage, 59(3):2142–2154, 2 2012.
[70] R. H. R. P uim e al. ICA-AROMA: A obus ICA-
based s a egy o emo ing mo ion a i ac s om MRI da a.
Neu oImage, 112:267–277, May 2015.
[71] C. Caballe o-Gaudes and R. C. Reynolds. Me hods o cleaning
he BOLD MRI signal. Neu oImage, 154:128–149, July 2017.
[72] D. Co des e al. F equencies con ibu ing o unc ional con-
nec i i y in he ce eb al co ex in ” es ing-s a e” da a. AJNR.
Ame ican jou nal o neu o adiology, 22(7):1326–1333, Augus
2001.
[73] O. Es eban e al. MRIP ep: a obus p ep ocessing pipeline
o unc ional MRI. Na u e Me hods, 16(1):111–116, Janua y
2019.
Bibliog aphy 113
[74] S. Whi ield-Gab ieli and A. Nie o-Cas anon. Conn : A Func-
ional Connec i i y Toolbox o Co ela ed and An ico ela ed
B ain Ne wo ks. B ain Connec i i y, 2(3):125–141, June 2012.
[75] C. Came on e al. Towa ds Au oma ed Analysis o Connec-
omes: The Con igu able Pipeline o he Analysis o Connec-
omes (C-PAC). F on ie s in Neu oin o ma ics, 7, 2013.
[76] S. Mo i and J. Zhang. P inciples o di usion enso imaging
and i s applica ions o basic neu oscience esea ch. Neu on,
51(5):527–539, Sep embe 2006.
[77] S. Mo i e al. Th ee-dimensional acking o axonal p ojec ions
in he b ain by magne ic esonance imaging. Annals o Neu o-
logy, 45(2):265–269, Feb ua y 1999.
[78] T. E. J. Beh ens e al. Cha ac e iza ion and p opaga ion o
unce ain y in di usion-weigh ed MR imaging. Magne ic Res-
onance in Medicine, 50(5):1077–1088, No embe 2003.
[79] J.-D. Tou nie e al. Robus de e mina ion o he ib e o i-
en a ion dis ibu ion in di usion MRI: Non-nega i i y con-
s ained supe - esol ed sphe ical decon olu ion. Neu oImage,
35(4):1459–1472, May 2007.
[80] J. Tou nie e al. Imp o ed p obabilis ic s eamlines ac o-
g aphy by 2nd o de in eg a ion o e ib e o ien a ion dis i-
bu ions. P oceedings o he In e na ional Socie y o Magne ic
Resonance in Medicine, 1670, 2010.
[81] B. T. T. Yeo e al. The o ganiza ion o he human ce eb al
co ex es ima ed by in insic unc ional connec i i y. Jou nal
o Neu ophysiology, 106(3):1125–1165, 2011.
[82] A. Schae e e al. Local-global pa cella ion o he human ce eb-
al co ex om in insic unc ional connec i i y m i. Ce eb-
al Co ex (New Yo k, N.Y.: 1991), 28(9):3095–3114, 9 2018.
PMID: 28981612 PMCID: PMC6095216.
114 Bibliog aphy
[83] R. S. Desikan e al. An au oma ed labeling sys em o subdi id-
ing he human ce eb al co ex on m i scans in o gy al based
egions o in e es . Neu oImage, 31(3):968–980, 2006.
[84] N. Tzou io-Mazoye e al. Au oma ed ana omical labeling o
ac i a ions in spm using a mac oscopic ana omical pa cella ion
o he mni m i single-subjec b ain. Neu oImage, 15(1):273–
289, 1 2002. PMID: 11771995.
[85] M. F. Glasse e al. A mul i-modal pa cella ion o human ce eb-
al co ex. Na u e, 536(7615):171–178, Augus 2016.
[86] I. Diez e al. A no el b ain pa i ion highligh s he modula
skele on sha ed by s uc u e and unc ion. Scien i ic Repo s,
5(1):10532, 2015.
[87] A. Jimenez-Ma in e al. Open da ase s and code o mul i-
scale ela ions on s uc u e, unc ion and neu o-gene ics in he
human b ain. bio Xi p ep in , aug 2023.
[88] K. J. F is on. Func ional and e ec i e connec i i y in neu oima-
ging: A syn hesis. Human B ain Mapping, 2(1-2):56–78, 1994.
[89] M. A. Koch e al. An in es iga ion o unc ional and ana omical
connec i i y using magne ic esonance imaging. Neu oImage,
16(1):241–250, 2002.
[90] C. J. Honey e al. Ne wo k s uc u e o ce eb al co ex shapes
unc ional connec i i y on mul iple ime scales. P oceedings
o he Na ional Academy o Sciences o he Uni ed S a es o
Ame ica, 104(24):10240–10245, June 2007.
[91] J. J. C o s and D. J. Higham. A weigh ed communicabili y
measu e applied o complex b ain ne wo ks. Jou nal o he
Royal Socie y, In e ace, 6(33):411–414, Ap il 2009.
[92] G. Deco e al. Key ole o coupling, delay, and noise in es -
ing b ain luc ua ions. P oceedings o he Na ional Academy o
Bibliog aphy 115
Sciences o he Uni ed S a es o Ame ica, 106(25):10302–10307,
June 2009.
[93] C. J. Honey e al. Can s uc u e p edic unc ion in he human
b ain? Neu oImage, 52(3):766–776, 2010.
[94] Y. Adachi e al. Func ional connec i i y be ween ana omic-
ally unconnec ed a eas is shaped by collec i e ne wo k-le el e -
ec s in he macaque co ex. Ce eb al Co ex (New Yo k, N.Y.:
1991), 22(7):1586–1592, July 2012.
[95] H.-J. Pa k and K. F is on. S uc u al and unc ional b ain
ne wo ks: om connec ions o cogni ion. Science (New Yo k,
N.Y.), 342(6158):1238411, No embe 2013.
[96] J. Go˜ni e al. Res ing-b ain unc ional connec i i y p edic ed
by analy ic measu es o ne wo k communica ion. P oceedings
o he Na ional Academy o Sciences o he Uni ed S a es o
Ame ica, 111(2):833–838, Janua y 2014.
[97] A. Mess´e e al. P edic ing unc ional connec i i y om s uc-
u al connec i i y ia compu a ional models using MRI: an ex-
ensi e compa ison s udy. Neu oImage, 111:65–75, May 2015.
[98] M. B eakspea . Dynamic models o la ge-scale b ain ac i i y.
Na u e Neu oscience, 20(3):340–352, Ma ch 2017.
[99] K. Bansal e al. Pe sonalized b ain ne wo k models o assessing
s uc u e- unc ion ela ionships. Cu en Opinion in Neu obi-
ology, 52:42–47, Oc obe 2018.
[100] B. V´azquez-Rod ´ıguez e al. G adien s o s uc u e- unc ion
e he ing ac oss neoco ex. P oceedings o he Na ional
Academy o Sciences o he Uni ed S a es o Ame ica,
116(42):21219–21227, Oc obe 2019.
[101] L. E. Su´a ez e al. Linking s uc u e and unc ion in mac oscale
b ain ne wo ks. T ends in Cogni i e Sciences, 24(4):302–315,
2020.
122 Bibliog aphy
[158] J. Lee and T. Coope . Pa hogenic mechanisms o myo onic
dys ophy. Biochemical Socie y T ansac ions, 37(6):1281–1286,
12 2009.
[159] J. D. Lueck e al. Chlo ide channelopa hy in myo onic dys-
ophy esul ing om loss o pos ansc ip ional egula ion
o clcn1. Ame ican Jou nal o Physiology-Cell Physiology,
292(4):C1291–C1297, 4 2007.
[160] M. San o o e al. Al e na i e splicing o human insulin ecep o
gene (ins ) in ype i and ype ii skele al muscle ibe s o pa ien s
wi h myo onic dys ophy ype 1 and ype 2. Molecula and
Cellula Biochemis y, 380(1-2):259–265, 8 2013.
[161] C. Fugie e al. Mis egula ed al e na i e splicing o bin1 is
associa ed wi h ubule al e a ions and muscle weakness in
myo onic dys ophy. Na u e Medicine, 17(6):720–725, 6 2011.
PMID: 21623381.
[162] A. Mankodi e al. Myo onic dys ophy in ansgenic mice ex-
p essing an expanded cug epea . Science (New Yo k, N.Y.),
289(5485):1769–1773, 9 2000. PMID: 10976074.
[163] A. Sis iaga e al. Cogni i e/pe sonali y pa e n and iple ex-
pansion size in adul myo onic dys ophy ype 1 (dm1): C g
epea s, cogni ion and pe sonali y in dm1. Psychological Medi-
cine, 40(3):487–495, 3 2010.
[164] S. Winblad e al. Cogni i e de ici s and c g epea expansion
size in classical myo onic dys ophy ype 1 (dm1). Beha io al
and b ain unc ions: BBF, 2:16, 5 2006. PMID: 16696870 PM-
CID: PMC1475858.
[165] C. Ca pen ie e al. Tau exon 2 esponsi e elemen s de egu-
la ed in myo onic dys ophy ype i a e p oximal o exon 2 and
syne gis ically egula ed by mbnl1 and mbnl2. Biochimica E
Biophysica Ac a, 1842(4):654–664, 4 2014. PMID: 24440524.

Bibliog aphy 123
[166] H. Jiang e al. Myo onic dys ophy ype 1 is associa ed wi h
nuclea oci o mu an na, seques a ion o muscleblind p o-
eins and de egula ed al e na i e splicing in neu ons. Hu-
man Molecula Gene ics, 13(24):3079–3088, 12 2004. PMID:
15496431.
[167] A. Annic e al. Ea ly dopasensi i e pa kinsonism ela ed o
myo onic dys ophy ype 2. Mo emen Diso de s: O icial
Jou nal o he Mo emen Diso de Socie y, 23(14):2100–2101,
10 2008. PMID: 18759330.
[168] K. I oh e al. Neu opa hology does no co ela e wi h egional
di e ences in he ex en o expansion o c g epea s in he b ain
wi h myo onic dys ophy ype 1. Ac a His ochemica E Cy o-
chemica, 43(6):149–156, 12 2010. PMID: 21245981 PMCID:
PMC3015052.
[169] L. Se a e al. How gene ics a ec s he b ain o p oduce highe -
le el dys unc ions in myo onic dys ophy ype 1. Func ional
Neu ology, 30(1):21–31, 3 2015. PMID: 26214024 PMCID:
PMC4520669.
[170] J. Ma hieu e al. Assessmen o a disease-speci ic muscula
impai men a ing scale in myo onic dys ophy. Neu ology,
56(3):336–340, 2 2001.
[171] D. Wechsle and N. Seisdedos Cube o. WAIS-III escala de in-
eligencia de Wechsle pa a adul os-III. TEA Ediciones, 1999.
OCLC: 981374932.
[172] M. D. Lezak e al. Neu opsychological assessmen , 4 h ed.
Neu opsychological assessmen , 4 h ed. Ox o d Uni e si y
P ess, New Yo k, NY, US, 2004. page: xi , 1016.
[173] M. Casals-Coll e al. Es udios no ma i os espa˜noles en po-
blaci´on adul a jo en (p oyec o neu ono ma j´o enes): no mas
pa a los es de luencia e bal. Neu olog´ıa, 28(1):33–40, 1 2013.
124 Bibliog aphy
[174] J. Pena-Casano a e al. Spanish mul icen e no ma i e s udies
(neu ono ma p ojec ): No ms o e bal luency es s. A chi es
o Clinical Neu opsychology, 24(4):395–411, 6 2009.
[175] A. Rey. FTes de Copia y Rep oducci´on de una Figu a Com-
pleja. Tea Ediciones, S.A., 2009.
[176] N. Adams. Con ibu ions o neu opsychological assessmen : A
clinical manual, 2nd ed. Neu ology, 45(8):1637–1637, 8 1995.
[177] G. Douaud e al. Ana omically ela ed g ey and whi e ma -
e abno mali ies in adolescen -onse schizoph enia. B ain,
130(9):2375–2386, 2007.
[178] C. D. Good e al. A oxel-based mo phome ic s udy o ageing
in 465 no mal adul human b ains. Neu oImage, 14(1 I):21–36,
2001.
[179] J. L. R. Ande sson e al. Non-linea egis a ion, aka spa ial
no malisa ion. m ib echnial epo 07ja2. Ox o d Cen e o
Func ional Magne ic Resonance Imaging o he B ain, Depa -
men o Clinical Neu ology, Ox o d Uni e si y, Ox o d, UK,
(June):22, 2007.
[180] S. M. Smi h e al. T ac -based spa ial s a is ics: oxelwise ana-
lysis o mul i-subjec di usion da a. Neu oImage, 31(4):1487–
1505, 7 2006. PMID: 16624579.
[181] S. M. Smi h and T. E. Nichols. Th eshold- ee clus e enhance-
men : add essing p oblems o smoo hing, h eshold dependence
and localisa ion in clus e in e ence. Neu oImage, 44(1):83–98,
1 2009. PMID: 18501637.
[182] J. A lo h e al. Re-anno a o : Anno a ion pipeline o mic oa -
ay p obe sequences. PLOS ONE, 10(10):e0139516, 10 2015.
[183] A. R. Lai d e al. B ainmap: he social e olu ion o a human
b ain mapping da abase. Neu oin o ma ics, 3(1):65–78, 2005.
PMID: 15897617.
Bibliog aphy 125
[184] S. B. Eickho e al. Coo dina e-based ac i a ion likelihood
es ima ion me a-analysis o neu oimaging da a: a andom-
e ec s app oach based on empi ical es ima es o spa ial unce -
ain y. Human B ain Mapping, 30(9):2907–2926, 9 2009. PMID:
19172646 PMCID: PMC2872071.
[185] J. B. Bu e al. Gene a i e modeling o b ain maps wi h spa ial
au oco ela ion. Neu oImage, 220:117038, 10 2020.
[186] J. Bux on e al. De ec ion o an uns able agmen o dna
speci ic o indi iduals wi h myo onic dys ophy. Na u e,
355(6360):547–548, 2 1992. PMID: 1346924.
[187] Y. H. Fu e al. An uns able iple epea in a gene ela ed
o myo onic muscula dys ophy. Science (New Yo k, N.Y.),
255(5049):1256–1258, 3 1992. PMID: 1546326.
[188] H. G. Ha ley e al. Expansion o an uns able dna egion
and pheno ypic a ia ion in myo onic dys ophy. Na u e,
355(6360):545–546, 2 1992. PMID: 1346923.
[189] M. Mahade an e al. Myo onic dys ophy mu a ion: an un-
s able c g epea in he 3’ un ansla ed egion o he gene. Sci-
ence (New Yo k, N.Y.), 255(5049):1253–1255, 3 1992. PMID:
1546325.
[190] A. Suzuki e al. Mkbp, a no el membe o he small hea shock
p o ein amily, binds and ac i a es he myo onic dys ophy p o-
ein kinase. The Jou nal o Cell Biology, 140(5):1113–1124, 3
1998. PMID: 9490724 PMCID: PMC2132705.
[191] R. S. Sa ku e al. Abe an egula ion o insulin ecep o
al e na i e splicing is associa ed wi h insulin esis ance in
myo onic dys ophy. Na u e Gene ics, 29(1):40–47, 9 2001.
PMID: 11528389.
[192] J. Shin e al. Impai ed neu ode elopmen by he low complexi y
domain o cpeb4 e eals a con e gen pa hway wi h neu ode-
126 Bibliog aphy
gene a ion. Scien i ic Repo s, 6:29395, 2016. PMID: 27381259
PMCID: PMC4933966.
[193] G. Ayalon e al. An anky in-based mechanism o unc ional
o ganiza ion o dys ophin and dys oglycan. Cell, 135(7):1189–
1200, 12 2008. PMID: 19109891.
[194] K.-Y. Lee e al. Compound loss o muscleblind-like unc ion in
myo onic dys ophy. EMBO molecula medicine, 5(12):1887–
1900, 12 2013. PMID: 24293317 PMCID: PMC3914532.
[195] M. Nakamo i e al. Splicing bioma ke s o disease se e i y in
myo onic dys ophy. Annals o Neu ology, 74(6):862–872, 12
2013. PMID: 23929620 PMCID: PMC4099006.
[196] A. Woj kowiak-Szlachcic e al. Sho an isense-locked nuc-
leic acids (all-lnas) co ec al e na i e splicing abno mali ies in
myo onic dys ophy. Nucleic Acids Resea ch, 43(6):3318–3331,
3 2015. PMID: 25753670 PMCID: PMC4381072.
[197] W. Dansi hong e al. Mbnl1 is he p ima y de e minan o
ocus o ma ion and abe an insulin ecep o splicing in dm1.
The Jou nal o Biological Chemis y, 280(7):5773–5780, 2 2005.
PMID: 15546872.
[198] R. N. Kanadia e al. A muscleblind knockou model
o myo onic dys ophy. Science (New Yo k, N.Y.),
302(5652):1978–1980, 12 2003. PMID: 14671308.
[199] M. Nakamo i e al. Abe an myokine signaling in congeni al
myo onic dys ophy. Cell Repo s, 21(5):1240–1252, 10 2017.
PMID: 29091763 PMCID: PMC5689469.
[200] H. Du e al. Abe an al e na i e splicing and ex acellula
ma ix gene exp ession in mouse models o myo onic dys ophy.
Na u e S uc u al & Molecula Biology, 17(2):187–193, 2 2010.
PMID: 20098426 PMCID: PMC2852634.
Bibliog aphy 127
[201] M. Hao e al. Muscleblind-like 2 (mbnl2) -de icien mice as a
model o myo onic dys ophy. De elopmen al Dynamics: An
O icial Publica ion o he Ame ican Associa ion o Ana omis s,
237(2):403–410, 2 2008. PMID: 18213585.
[202] J. E. Mo gan e al. Nec op osis media es myo ib e dea h in
dys ophin-de icien mice. Na u e Communica ions, 9(1):3655,
2018. PMID: 30194302 PMCID: PMC6128848.
[203] M. Wehling-Hen icks e al. Klo ho gene silencing p omo es
pa hology in he mdx mouse model o duchenne muscula dys-
ophy. Human Molecula Gene ics, 25(12):2465–2482, 2016.
PMID: 27154199 PMCID: PMC5181628.
[204] M. Wehling-Hen icks e al. Mac ophages escape klo ho gene
silencing in he mdx mouse model o duchenne muscula dys-
ophy and p omo e muscle g ow h and inc ease sa elli e cell
numbe s h ough a klo ho-media ed pa hway. Human Molecu-
la Gene ics, 27(1):14–29, 2018. PMID: 29040534 PMCID:
PMC5886268.
[205] G. T. Co be e al. P p is a cen al playe in oxici y medi-
a ed by soluble agg ega es o neu odegene a ion-causing p o-
eins. Ac a Neu opa hologica, 139(3):503–526, 3 2020.
[206] E. Kabashi e al. Ta dbp mu a ions in indi iduals wi h spo adic
and amilial amyo ophic la e al scle osis. Na u e Gene ics,
40(5):572–574, 5 2008. PMID: 18372902.
[207] Z. Simmons. Wha ’s in he li e a u e? Jou nal o Clinical Neu -
omuscula Disease, 10(4):202–207, 6 2009. PMID: 19494733.
[208] C. A. Bouche e al. A no el homeodomain-encoding gene is
associa ed wi h a la ge cpg island in e up ed by he myo onic
dys ophy uns able (c g)n epea . Human Molecula Gene ics,
4(10):1919–1925, 10 1995. PMID: 8595416.

128 Bibliog aphy
[209] S. K. Hea h e al. Cha ac e isa ion o exp ession o mdmahp, a
homeodomain-encoding gene a he mu ine dm locus. Human
Molecula Gene ics, 6(5):651–657, 5 1997. PMID: 9158137.
[210] J. P. O engo e al. Expanded c g epea s wi hin he dmpk
3’ u causes se e e skele al muscle was ing in an inducible
mouse model o myo onic dys ophy. P oceedings o he Na-
ional Academy o Sciences o he Uni ed S a es o Ame -
ica, 105(7):2646–2651, 2 2008. PMID: 18272483 PMCID:
PMC2268190.
[211] Y. Kino e al. Mbnl and cel p o eins egula e al e na i e spli-
cing o he skele al muscle chlo ide channel clcn1. Nucleic Acids
Resea ch, 37(19):6477–6490, 10 2009. PMID: 19720736 PMCID:
PMC2770659.
[212] N. B esolin e al. Cogni i e impai men in duchenne muscula
dys ophy. Neu omuscula Diso de s, 4(4):359–369, 7 1994.
[213] J. M. E as i e al. De iciency o a glycop o ein compon-
en o he dys ophin complex in dys ophic muscle. Na u e,
345(6273):315–319, 5 1990. PMID: 2188135.
[214] M. Naidoo and K. An hony. Dys ophin dp71 and he neu o-
pa hophysiology o duchenne muscula dys ophy. Molecula
Neu obiology, 57(3):1748–1767, 3 2020.
[215] M. Thanga ajh e al. Rela ionships be ween DMD mu a ions
and neu ode elopmen in dys ophinopa hy. Neu ology, pp.
10.1212/WNL.0000000000008363, 10 2019.
[216] R. Wo on. Muscula dys ophies: diseases o he
dys ophin-glycop o ein complex. Science (New Yo k, N.Y.),
270(5237):755–756, 11 1995. PMID: 7481760.
[217] L. Machuca-Tzili e al. Flies de icien in muscleblind p o ein
model ea u es o myo onic dys ophy wi h al e ed splice o ms
o z-band associa ed ansc ip s. Human Gene ics, 120(4):487–
499, 11 2006. PMID: 16927100.
Bibliog aphy 129
[218] X. Lin e al. Failu e o mbnl1-dependen pos -na al splicing
ansi ions in myo onic dys ophy. Human Molecula Gene ics,
15(13):2087–2097, 7 2006. PMID: 16717059.
[219] Y. Yamashi a e al. Fou pa ame e s inc ease he sensi i i y
and speci ici y o he exon a ay analysis and disclose 25 no el
abe an ly spliced exons in myo onic dys ophy. Jou nal o
Human Gene ics, 57(6):368–374, 6 2012. PMID: 22513715.
[220] Z. Z. Tang e al. Muscle weakness in myo onic dys ophy asso-
cia ed wi h mis egula ed splicing and al e ed ga ing o ca( )1.1
calcium channel. Human Molecula Gene ics, 21(6):1312–1324,
3 2012. PMID: 22140091 PMCID: PMC3284119.
[221] A. Fab ega e al. Reac ome pa hway analysis: a high-
pe o mance in-memo y app oach. BMC Bioin o ma ics,
18(1):142, 12 2017.
[222] J. Ri chie e al. T ansc ip omic cha ac e iza ion o m i con-
as wi h ocus on he 1-w/ 2-w a io in he ce eb al co ex.
Neu oImage, 174:504–517, 7 2018.
[223] A. Fo ni o e al. B idging he gap be ween connec ome and
ansc ip ome. T ends in Cogni i e Sciences, 23(1):34–50, 1
2019.
[224] R. Rome o-Ga cia e al. S uc u al co a iance ne wo ks a e
coupled o exp ession o genes en iched in sup ag anula laye s
o he human co ex. Neu oImage, 171:256–267, 5 2018.
[225] R. Rome o-Ga cia e al. Synap ic and ansc ip ionally down-
egula ed genes a e associa ed wi h co ical hickness di e -
ences in au ism. Molecula Psychia y, 24(7):1053–1064, 7 2019.
[226] V. Tam e al. Bene i s and limi a ions o genome-wide associ-
a ion s udies. Na u e Re iews Gene ics, 20(8):467–484, 8 2019.
[227] N. Se gean e al. Dys egula ion o human b ain mic o ubule-
associa ed au m na ma u a ion in myo onic dys ophy ype 1.
130 Bibliog aphy
Human Molecula Gene ics, 10(19):2143–2155, 9 2001. PMID:
11590131.
[228] P. Ve me sch e al. Speci ic au a ian s in he b ains o pa ien s
wi h myo onic dys ophy. Neu ology, 47(3):711–717, 9 1996.
PMID: 8797469.
[229] O. Le oy e al. B ain-speci ic change in al e na i e splicing
o au exon 6 in myo onic dys ophy ype 1. Biochimica E
Biophysica Ac a, 1762(4):460–467, 4 2006. PMID: 16487687.
[230] C. M. Dhaenens e al. Mis-splicing o au exon 10 in myo onic
dys ophy ype 1 is ep oduced by o e exp ession o cel 2
bu no by mbnl1 silencing. Biochimica E Biophysica Ac a,
1812(7):732–742, 7 2011. PMID: 21439371.
[231] D. Ghanem e al. Al e ed splicing o au in dm1 is di e en
om he oe al splicing p ocess. FEBS le e s, 583(4):675–679,
2 2009. PMID: 19166838.
[232] R. C. Flo ek e al. Elec oca diog aphic abno mali ies in
pa ien s wi h myo onic dys ophy. The Wes e n Jou nal o
Medicine, 153(1):24–27, 7 1990. PMID: 2202157 PMCID:
PMC1002461.
[233] P. V. F agola e al. Ca diac in ol emen in myo onic dys-
ophy. The Ame ican Jou nal o Ca diology, 74(10):1070–1072,
11 1994. PMID: 7977052.
[234] J. D. Rhodes e al. Ac i a ion o he inna e immune esponse
and in e e on signalling in myo onic dys ophy ype 1 and ype
2 ca a ac s. Human Molecula Gene ics, 21(4):852–862, 2 2012.
[235] M. Nakamo i e al. Abe an ly spliced -dys ob e in al e s -
syn ophin binding in myo onic dys ophy ype 1. Neu ology,
70(9):677–685, 2 2008.
Bibliog aphy 131
[236] A. Kiuchi e al. P esenile appea ance o abundan alzheime ’s
neu o ib illa y angles wi hou senile plaques in he b ain in
myo onic dys ophy. Ac a Neu opa hologica, 82(1):1–5, 6 1991.
[237] O. He n´andez-He n´andez e al. Myo onic dys ophy c g ex-
pansion a ec s synap ic esicle p o eins, neu o ansmission and
mouse beha iou . B ain, 136(3):957–970, 3 2013.
[238] O. He n´andez-He n´andez e al. Synap ic p o ein dys egula ion
in myo onic dys ophy ype 1: Disease neu opa hogenesis bey-
ond missplicing. Ra e Diseases, 1(1):e25553, 1 2013.
[239] H. Hodges e al. Au ism spec um diso de : de ini ion, epidemi-
ology, causes, and clinical e alua ion. T ansla ional Pedia ics,
9(S1):S55–S65, 2 2020.
[240] M.-C. Lai e al. Au ism. Lance (London, England),
383(9920):896–910, 3 2014. PMID: 24074734.
[241] A. Masi e al. An o e iew o au ism spec um diso de ,
he e ogenei y and ea men op ions. Neu oscience Bulle in,
33(2):183–193, 4 2017.
[242] M. H. Mehling and M. J. Tass´e. Se e i y o au ism spec um
diso de s: Cu en concep ualiza ion, and ansi ion o dsm-5.
Jou nal o Au ism and De elopmen al Diso de s, 46(6):2000–
2016, 6 2016.
[243] R. Bhanda i e al. Neu opsychopa hology o au ism spec um
diso de : Complex in e play o gene ic, epigene ic, and en i -
onmen al ac o s. In M. M. Essa and M. W. Qo on leh, edi o s,
Pe sonalized Food In e en ion and The apy o Au ism Spec-
um Diso de Managemen , olume 24, pp. 97–141. Sp inge
In e na ional Publishing, Cham, 2020. collec ion- i le: Ad-
ances in Neu obiology DOI: 10.1007/978-3-030-30402-7 4.
[244] V. Eapen. Gene ic basis o au ism: is he e a way o wa d?:.
Cu en Opinion in Psychia y, 24(3):226–236, 5 2011.