Sahana e al. Gene Sel E ol (2016) 48:35
DOI 10.1186/s12711-016-0215-z
RESEARCH ARTICLE
A 0.5-Mbp dele ion onbo ine
ch omosome 23 is a s ong candida e
o s illbi h inNo dic Red ca le
Gou am Sahana1* , Te hi Iso‑Tou u2, Xiaoping Wu1, Ul ik Sande Nielsen3, Di k‑Jan de Koning4,
Mogens Sandø Lund1, Johanna Vilkki2 and Be n Guldb and sen1
Abs ac
Backg ound: A whole‑genome associa ion s udy o 4631 p ogeny‑ es ed No dic Red dai y ca le bulls using
impu ed nex ‑gene a ion sequencing da a e ealed a majo quan i a i e ai locus (QTL) ha a ec s bi h index (BI)
on Bos au us au osome (BTA) 23. We analyzed his QTL o iden i y which o he componen ai s o BI a e a ec ed
and unde s and i s molecula basis.
Resul s: A genome‑wide scan o BI in No dic Red dai y ca le de ec ed majo QTL on BTA6, 14 and 23. The s onges
associa ed single nucleo ide polymo phism (SNP) on BTA23 was loca ed a 13,313,896 bp wi h
−log10(p)
=
50.63
.
Analyses o componen ai s showed ha he QTL had a la ge e ec on s illbi h. Based on he 10 mos s ongly asso‑
cia ed SNPs wi h s illbi h, we cons uc ed a haplo ype. Among his haplo ype’s alleles, HAPQTL had a la ge nega i e
e ec on s illbi h. No animals we e ound o be homozygous o HAPQTL. Analysis o s illbi h eco ds ha we e ca ‑
ego ized by ca ie s a us o HAPQTL o he si e and ma e nal g andsi e sugges ed ha his haplo ype had a ecessi e
mode o inhe i ance. Illumina Bo ineHD BeadChip geno ypes and geno ype in ensi y da a indica ed a ch omosomal
dele ion be ween 12.28 and 12.81 Mbp on BTA23. An independen se o Illumina Bo ine50k BeadChip geno ypes
iden i ied a ecessi e le hal haplo ype ha spanned he dele ed egion.
Conclusions: A dele ed egion o app oxima ely 500 kb ha spans h ee genes on BTA23 was iden i ied and is a
s ong candida e QTL wi h a la ge e ec on BI by inc easing s illbi h.
© 2016 Sahana e al. This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional License
(h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium,
p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license,
and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e (h p://c ea i ecommons.o g/
publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed.
Backg ound
P e iously, he molecula causes o gene ic de ec s in
ca le we e mainly de ec ed by examining dead o mal-
o med cal es [1–9]. The ecen a ailabili y o high-
h oughpu geno yping pla o ms and nex -gene a ion
sequencing echnologies has subs an ially accele a ed
he disco e y o le hal gene ic ac o s [10–14]. How-
e e , polymo phisms ha a e no associa ed wi h cha -
ac e is ic pheno ypic mani es a ions, such as physically
mal o med cal es, emain di icul o iden i y. A comple-
men a y app oach is quan i a i e ai locus (QTL) map-
ping. Fo example, a QTL wi h a la ge e ec on e ili y
in No dic Red dai y ca le was i s epo ed on Bos au-
us au osome (BTA) 12 [15]. The molecula basis o his
QTL was subsequen ly iden i ied as a 660-kb dele ion
ha causes emb yonic dea h [16]. A whole-genome asso-
cia ion scan in No dic Red dai y ca le de ec ed a majo
QTL o bi h index (BI) on BTA23. BI is a composi e
index ha desc ibes he con ibu ion o addi i e gene ic
e ec s o a cal ’s gene ic po en ial o be bo n.
In his s udy, we analyzed wo componen ai s ha
con ibu e o BI (i.e. s illbi h and cal ing ease) o he
QTL on BTA23. Ou objec i es we e: (1) o iden i y
which componen ai s a e a ec ed by he BI QTL on
BTA23, and (2) o unde s and i s molecula basis.
Open Access
G
ene ics
S
elec ion
E olu ion
*Co espondence: gou [email protected]
1 Depa men o Molecula Biology and Gene ics, Cen e o Quan i a i e
Gene ics and Genomics, Aa hus Uni e si y, 8830 Tjele, Denma k
Full lis o au ho in o ma ion is a ailable a he end o he a icle
Page 2 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
Me hods
Animal Ca e and Use Commi ee app o al was no
equi ed o his s udy because he pheno ype and geno-
ype da a used we e ou inely collec ed as pa o a b eeding
p og am, and no li e animal expe imen s we e pe o med.
Animals and ai s
The s udy was ca ied ou on h ee popula ions o No dic
Red dai y ca le (RDC) om Denma k (RDCDNK), Fin-
land (RDCFIN) and Sweden (RDCSWE). A o al o 4631
p ogeny- es ed bulls wi h b eeding alues o BI and i s
componen ai s and geno ype in o ma ion we e used
o QTL mapping.
As dependen ai s, we analyzed he de- eg essed
b eeding alues (DRP) o cal ing- ela ed ai s. We
scanned he b eeding alues o BI o iden i y QTL. BI is
a compound index ha p edic s a si e’s o al di ec addi-
i e gene ic e ec on cal ing by combining di ec (cal )
e ec s o s illbi h (SB) in he i s (SBF) and la e (SBL)
cal ings and cal ing ease (CE) in he i s (CEF) and la e
(CEL) cal ings. Cal size (CS) is no included in he p e-
dic ion o BI by he No dic Ca le Gene ic E alua ion
(NAV; h p://www.no diceb .in o). Howe e , bi h weigh
is posi i ely co ela ed wi h cal ing assis ance/di icul y
[17]. The e o e, we analyzed CS in he i s (CSF) and la e
(CSL) cal ings. Cal es ha we e ali e o dead wi hin 24h
a e bi h we e eco ded as 1 o 0 ( o SB), espec i ely.
Fa me s subjec i ely assessed and classi ied CE in o wo
ca ego ies in Sweden and ou ca ego ies in Denma k and
Finland) and CS in o ou ca ego ies bu his was eco ded
only in Denma k. We analyzed de- eg essed b eeding al-
ues [18] ha we e p oduced as pa o he ou ine b eed-
ing alue e alua ion by he NAV. Fo u he in o ma ion
on he eco ding and gene ic e alua ion o di ec cal ing
ai s, see [19] and h p://www.no diceb .in o.
Geno yping o bulls
All bulls we e geno yped by using he Bo ineSNP50 Bead-
Chip (Illumina, San Diego, CA) e sion 1 o 2 and genomic
DNA ex ac ed om whole blood o semen. Quali y
con ol (QC) o selec single nucleo ide polymo phisms
(SNPs) we e as ollows: a minimum call a e o 85% o
samples, and exclusion o SNPs wi h a call a e less han
95%, a mino allele equency (MAF) less han 0.5% and
a signi ican de ia ion om Ha dy–Weinbe g p opo ions
(HWP; p<10−5). A e QC, 43,415 SNPs we e e ained o
analyses. Genomic posi ions o he SNPs we e based on
he UMD3.1 Bo ine Genome Assembly [20].
Impu a ion ohigh‑densi y (HD) and ull‑genome
sequence
The 50 k SNPs we e impu ed o ull sequence in wo
s eps. Fi s , 50k geno ypes o bulls we e impu ed o HD
geno ypes by using he IMPUTE2 so wa e wi h de aul
pa ame e s, excep o he e ec i e popula ion size
(Ne)=100 [21]. The e e ence popula ion wi h HD geno-
ypes consis ed o 2036 bulls (902 Hols ein, 735 No dic
Red and 399 Danish Je sey). The same QC pa ame e s
ha we e used o he 50k chip we e used o HD da a.
A e QC, 648,219 SNPs emained o he HD chip.
In he second impu a ion s ep, using whole-genome
sequencing (WGS) da a om 242 dai y ca le as a e e -
ence, 12,322 bulls om h ee b eeds (6032 Hols ein, 1645
Je sey and 4645 No dic Red) impu ed o HD geno ypes
we e u he impu ed o he ull-sequence le el by using
Beagle so wa e [22]. Sequences o he e e ence popu-
la ion used o impu ed animals consis ed o WGS da a
ha we e ob ained by Aa hus Uni e si y [23, 24] and by
he 1000 Bull Genome P ojec un2 [14]. Ch omosomes
we e di ided in o windows o abou 20,000 consecu i e
SNPs, wi h an o e lap o 250 SNPs a each end o mini-
mize impu a ion e o s a each end o he windows. Fo
de ails on nex -gene a ion sequencing da a and impu a-
ion s eps, see [23].
Beagle 3.3.2 [22] was used o p e-phase e e ence
da a and o impu e om impu ed HD geno ypes o he
ull-sequence a ian s. All SNPs wi h an impu a ion ce -
ain y (R2) alue less han 0.9 we e emo ed. A o al o
8,938,927 SNPs emained ha we e dis ibu ed ac oss
he 29 bo ine au osomes [25].
S a is ical models o associa ion analysis
Fi s , a WGS scan was pe o med using a si e model
wi hou conside ing ela ionships be ween indi iduals,
excep o he si es and hei sons. Based on he esul s
om his scan, we selec ed a egion on BTA23 o u he
analysis wi h an animal model ha conside ed ela ion-
ships be ween all indi iduals, using bo h single-ma ke
and haplo ype-based analyses.
Si e model o  he whole‑genome scan
A SNP-by-SNP analysis was ca ied ou , in which each
SNP was es ed o associa ion wi h he pheno ype (de-
eg essed b eeding alue). The ollowing linea mixed
model (LMM) was used o es ima e SNP e ec s:
whe e yij is he de- eg essed b eeding alue o he j h son
om he hal -sib (si e) amily i; μ is he gene al mean; b
is he allelic subs i u ion e ec ; and xij ( anging om 0
o 2) is he allelic dose o he j h indi idual o he SNP
ob ained by impu a ion. si is he andom e ec o he
i h hal -sib amily, assumed o be no mally dis ibu ed
as
si
∼N

0, σ
2
s
, whe e σs
2 is he si e a iance. eij is he
andom esidual o son j om he hal -sib amily i and is
assumed o be no mally dis ibu ed as
e
∼
N
(0
,
W
−1
σ
2
e
),
yij
=µ+
bxij
+
si
+
eij,
Page 3 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
whe e
e
is he ec o o andom esiduals (ei);
σ2
e
is he
e o a iance; and
W
is a diagonal ma ix whe e he
diagonal elemen s a e weigh s o he DRP. The weigh o
he i h animal was es ima ed by wi= i
2/(1− i
2), whe e
i
2 is he eliabili y o he ij h animal’s DRP. Values o i
2
ha we e highe han 0.98 we e se o 0.98 o a oid ha -
ing excessi ely la ge weigh s o si es wi h la ge numbe s
o p ogeny eco ds.
All s a is ical analyses we e conduc ed by using he
so wa e DMU [26]. The null hypo hesis H0:
b=0
was
es ed wi h a wo-sided es . A SNP was conside ed o
ha e a signi ican associa ion wi h a ai i he
−log10(p)
was highe han 8.25 (a e Bon e oni mul iple- es ing
co ec ion o 8,938,927 simul aneous es s), co e-
sponding o a nominal genome-wide signi icance le el o
p
=
0.05
.
LMM analysis o  he a ge egion onBTA23
LMM analysis was used o associa ion analyses o he
a ge egion on BTA23 o BI and i s componen ai s
(Table1). Associa ion be ween he SNP and pheno ype
was assessed by a single-locus eg ession analysis o he
pheno ype on o he allelic dosage o each SNP sepa-
a ely, using an LMM [27] as ollows:
whe e yj is he pheno ype (DRP) o he j h bull; μ is he
o e all mean; b is he allelic subs i u ion e ec ; and xj
( anging om 0 o 2) is he allelic dose o he j h indi-
idual o he SNP. uj is he andom polygenic e ec wi h
a join mul i a ia e no mal dis ibu ion,
u
∼N

0,Aσ
2
u
,
whe e u is he ec o o polygenic e ec s (ui);
A
is he
addi i e gene ic ela ionship ma ix; and
σ2
u
is he poly-
genic a iance. ej is he andom esidual o he j h ani-
mal dis ibu ed as desc ibed in he si e model.
yj
=
µ
+
bxj
+
uj
+
ej,
The model was i ed by es ic ed maximum likelihood
(REML) using he so wa e DMU [26], which p o ided
es ima es o he ixed e ec s and hei s anda d e o s.
Tes ing o he e ec o a ma ke was done by using a
wo-sided - es agains a null hypo hesis o H0:
b=0
,
wi h a Bon e oni co ec ion o 8,938,927 mul iple es -
ing (
−log10
(
p
)=
8.25
) simila ly o he si e model.
Random haplo ype model
Haplo ypes o he 10 SNPs ha we e mos signi ican ly
associa ed in he LMM analyses and loca ed be ween
13,259,463 and 13,354,932bp on BTA23 we e ex ac ed
om Beagle’s WGS impu a ion ou pu  (Table 2). We
used a LMM ha included andom polygenic e ec s and
andom e ec s o he haplo ype, i.e. andom haplo ype
model (RHM) ollowing [28]. The RHM model was:
whe e qh1j and qh2j a e andom e ec s o he wo hap-
lo ypes (one each om he si e and dam) ca ied by
he j h indi idual, assumed o be no mally dis ibu ed
as
qhij
∼N

0, σ
2
h
, whe e σh
2 is he a iance o haplo-
ype e ec s. O he e ms in he RHM we e he same as
desc ibed o he LMM.
The signi icance o he haplo ype subs i u ion e ec
was assessed by a likelihood a io es ha compa ed he
RHM model o a null model ha included mean, poly-
genic e ec and andom e o e ms, bu no haplo ype
e ec s. Unde H0, he es s a is ic has a χ2 dis ibu-
ion wi h one deg ee o eedom. The analysis was pe -
o med in he DMU so wa e package [26]. A haplo ype
wi h a la ge nega i e p edic ed e ec on BI, i.e. HAPQTL
was iden i ied as he haplo ype ha ca ied he pu a i e
causal polymo phism o he QTL. The abo e LMM anal-
ysis was epea ed wi h indi idual coun s o HAPQTL ha
we e added as a ixed eg ession e ec . Ca ie s a us o
each bull wi h espec o HAPQTL was p edic ed o u -
he analysis, as desc ibed below.
Analysis o cal su i al asa unc ion o haplo ype ca ie
s a us o  he si e andma e nal g andsi e
Reco ded p egnancies we e classi ied in o ypes acco d-
ing o he ca ie s a us o he si e and he ma e nal
g andsi e o HAPQTL. Fo each p egnancy, we eco ded
whe he he cal was ali e o dead 24h a e bi h. A
ecessi e le hal mu a ion can be es ed by compa ing
ou comes om ma ing ypes, because he e is a 25%
p obabili y ha a p egnancy o which bo h he si e and
dam a e ca ie s will esul in a ec ed cal es. A LMM
was applied o es he e ec o ma ing ype on cal su -
i al a bi h.
Fou classes o ma ings we e de ined acco ding o he
HAPQTL ca ie s a us, i.e. (I) nonca ie si e ma ed o
yj=µ+qh1j+qh2j+uj+ej,
Table 1 Mos signi ican ly associa ed SNPs wi h di ec
cal ing ai s ob ained omlinea mixed model analyses
o  he a ge egion onch omosome 23
Su ix F and L a e o i s and la e cal ings
E ec s and SE a e gi en in b eeding index uni s
BI bi h index, SB s illbi h, CE cal ing ease, CS cal size, MAF mino allele
equency, SE s anda d e o
T ai Top associa ed
SNP MAF Allele
subs i u ion
e ec
SE p alue
BI Ch 23:13,313,896 0.059 −8.76 0.45 4.79e−80
SBF Ch 23:13,313,896 0.059 −7.35 0.45 8.09e−58
SBL Ch 23:13,313,896 0.059 −10.53 0.44 1.00e−120
CSF Ch 23:17,631,540 0.160 1.55 0.19 1.17e−16
CSL Ch 23:17,631,540 0.160 1.60 0.18 6.27e−19
CEF Ch 23:17,631,540 0.160 −1.86 0.21 3.34e−19
CEL Ch 23:17,581,154 0.145 −2.52 0.24 1.26e−25
Page 4 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
he daugh e o a nonca ie ma e nal g andsi e; (II)
nonca ie si e ma ed o he daugh e o a ca ie ma e -
nal g andsi e; (III) ca ie si e ma ed o he daugh e
o a nonca ie ma e nal g andsi e; and (IV) ca ie
si e ma ed o he daugh e o a ca ie ma e nal g and-
si e. S illbi hs o hei e s ( i s cal ing) and cows (la e
cal ings) we e analyzed sepa a ely. A o al o 3,932,927
cal ing eco ds om Denma k, Finland and Sweden
we e analyzed, wi h 2,785,085 eco ds o ma ing ype I,
536,614 o ma ing ype II, 512,714 o ma ing ype III
and 98,514 o ma ing ype IV. Numbe s o eco ds in
each ca ego y a e in Table3.
The i ed mixed model included pa i y and mon h o
insemina ion (by yea ) as ixed e ec s and he ma e nal
g andsi e as a andom e ec :
whe e yijkl is an indica o o cal su i al (0 o su i al
and 1 o s illbi h); pj is he e ec o pa i y; k is he e ec
o mon h and yea o insemina ion; and ml is he e ec
o ma ing ype. Vec o
u
o andom g andsi e e ec s ui
is assumed o be no mally dis ibu ed
u
∼N

0,σ2
gAs

,
whe e
As
is he addi i e gene ic ela ionship among si es
o dams de i ed om he pedig ee.
e
is a ec o o an-
dom indi idual e o e ms
eijkl
and is assumed o be
no mally dis ibu ed
e
∼N

0,Iσ
2
e
. Assuming HWP, he
expec ed p opo ion o concep uses ha a e homozy-
gous o HAPQTL is equal o 1/4 o he p obabili y ha
bo h pa en s a e ca ie s, co esponding o 0, 0,
p
4
(
1
+
p)
,
and
(1+p)
4
(
2
+
p)
o ma ing ypes I, II, III, and IV, espec i ely,
whe e p is he equency o he causa i e allele [16].
Sea ch o ch omosomal dele ions wi hin he QTL
o s illbi h
A ecessi e le hal mu a ion can be caused by a ch omo-
somal dele ion [29, 30]. We sea ched o ch omosomal
dele ions a he QTL using wo app oaches: (1) de ia ion
om HWP, and (2) loss o geno ype in ensi y based on
SNP a ay da a.
De ia ion omHWP
I he ecessi e le hal e ec is due o a ch omosomal
dele ion, hen an excess o homozygo es will be called
o he SNPs ha a e loca ed wi hin he dele ed egion
[16]. Impu ed HD SNP geno ypes in he QTL egion
we e in es iga ed o de ia ions om HWP. Among he
impu ed SNP geno ypes, we es ed o de ia ions o hei
equencies om HWP by using a χ2 dis ibu ion wi h
one deg ee o eedom. Fo SNPs ha de ia ed om
HWP, we in es iga ed whe he he e was an excess o a
de ici o he e ozygo es.
yijkl =µ+pj+ k+ml+ui+eijkl,
Loss o geno ype in ensi y
We had access o geno ype in ensi y da a o 243
RDCFIN bulls ha had been geno yped wi h he Illu-
mina Bo ineHD Geno yping BeadChip ha included
725,293 au osomal SNPs [16]. We isually inspec ed
he signal in ensi ies a ound he QTL egion using he
SVS8 p og am (Golden Helix) o geno yped bulls.
Reduced in ensi ies we e obse ed be ween SNP
Bo ineHD2300003056 (23:12,289,281, s109948445)
and SNP BOVINEHD2300003186 (23: 12,816,660,
s132859742). An animal was conside ed o ca y he
dele ion i i me wo condi ions: he a e age signal in en-
si y (‘Log2R a io’) was less han 0, and 97% o he SNPs
we e homozygous wi hin he egion de ined by he sig-
nal in ensi ies. Si es we e ca ego ized as ca ie s o non-
ca ie s o he ch omosomal dele ion (CHRDEL), and he
a e age in ensi y pe SNP was calcula ed o bo h g oups.
An indi idual ha ca ies CHRDEL will be e oneously
geno yped as homozygous o all loci wi hin he dele ed
egion because only one allele is p esen . Thus, a small
abe a ion (3%) was allowed, o accoun o possible gen-
o yping e o s.
Sea ch o  ecessi e le hal haplo ypes (absence
o homozygo es) using 50k geno ype da a
Absence o homozygo es o a common haplo ype among
li e indi iduals is s ong e idence o he p esence o a
ecessi e le hal allele [10]. Based on his, we sea ched o
a ecessi e le hal allele on BTA23 seg ega ing in No dic
Red dai y ca le. A o al o 19,309 No dic Red dai y ca -
le animals wi h 50k geno ypes, including 1119 SNPs on
BTA23, we e a ailable. We disca ded 136 SNPs ha had
a MAF less han 0.01 and 73 SNPs ha had e y s ong
de ia ions om HWP (p < 10−5). To a oid he in lu-
ence o a dele ion on phasing accu acy, we emo ed ou
SNPs ha we e loca ed wi hin he pu a i e dele ed egion
(be ween 12.2 and 12.9Mb).
Nex , we used Beagle 4.0 [31] o impu e spo adic
missing geno ypes and o in e he haplo ype phase.
Finally, 901 SNPs emained on BTA23. To iden i y a
haplo ype ha agged he dele ed egion, we selec ed a
10-SNP egion ha spanned he pu a i e dele ed egion
( i e SNPs ups eam and i e SNPs downs eam) and
co e ed a egion be ween posi ions 11,818,761 and
13,122,318bp. I was expec ed ha one o mo e haplo-
ypes in his window would be in igh linkage disequilib-
ium wi h he dele ed egion. I he dele ion is ecessi e
le hal, hen he haplo ype(s) ha ag(s) i should only be
p esen in he he e ozygous condi ion; i.e. no li e animal
homozygous o he ecessi e le hal haplo ype should be
ound.
Page 5 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
Resul s
Genome scan o bi h index
Figu e1 p esen s he Manha an plo o he QTL scan
o BI based on WGS da a in No dic Red dai y ca le.
Addi ional ile1: Table S1 p o ides he op associa ed
genome-wide signi ican SNPs o each ch omosome.
S ong associa ion signals o BI we e obse ed on BTA6,
14 and 23. P e iously, we epo ed a QTL o BI on BTA6
which inc eased cal size a bi h and adul s a u e [32]
and we a ibu ed he inc eases in cal ing di icul ies and
s illbi h o he inc eased cal size a bi h. In he p esen
s udy, we ocused on he QTL o BI ha is loca ed on
BTA23. The mos s ongly associa ed SNP was loca ed a
13,313,896bp ( s722178836) and had an allele subs i u-
ion e ec o -0.72 addi i e gene ic s anda d de ia ions
and
−log10 (p)
=
50.63
(Fig.2).
LMM analysis o  he a ge ed egion
BI and h ee cal ing ai s (SB, CE and CS) we e analyzed
o associa ion wi h SNPs in he a ge egion on BTA23
using a LMM. The SNPs ha we e mos s ongly associ-
a ed wi h each ai a e in Table1. The s onges asso-
cia ion signal was obse ed o SBL (ligh g ay do s in
Fig.3), ollowed by BI (ligh g ay do s in Fig.4) and SBF
(ligh g ay do s in Fig.5), wi h a peak a 13,313,896bp
( s722178836). Associa ion peaks o bo h CE and CS
we e posi ioned a 17.63 Mbp (see Addi ional ile2: Fig-
u es S1, S2). The
−log10 (p)
alues we e be ween 15.9 and
24.4, i.e. much smalle han hose obse ed o SB and BI.
Haplo ype‑based analysis o  he a ge ed egion
The s onges associa ion signal among he analyzed
ai s was ob ained o SBL (Table1; Fig.3). The 10 SNPs
ha we e mos s ongly associa ed wi h SBL we e used
o cons uc a haplo ype (Table 2). Di e si y o hese
10-SNP haplo ypes was ex emely low, wi h only i e o
Fig. 1 Manha an plo o single‑ a ian associa ion analysis o bi h index in No dic Red dai y ca le. Associa ion analyses we e ca ied ou using a
si e model o whole‑genome sequence a ian s
Fig. 2 Associa ion esul s o single‑ a ian analysis, ob ained using
a si e model, o bi h index on ch omosome 23 in No dic Red dai y
ca le. The X‑axis is ch omosomal loca ion in million base pai s and
Y‑axis is −log10(p)

Page 6 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
he 1024 possible haplo ypes being obse ed. Haplo ypes
and equencies o he No dic Red dai y ca le popula-
ion we e ATGTGAGTGA (0.9306), ATGTGAGTGC
(0.0059), GCACAGCGAA (0.0019), GCACAGCGAC
(0.0577) and GCACAGCGGA (0.0042).
When we analyzed haplo ype as a andom e ec in he
RHM, we obse ed ha he haplo ype GCACAGCGAC
educed he b eeding alues o SBL by 0.74 addi i e
gene ic s anda d de ia ions. The equency o ca ie s o
his haplo ype (HAPQTL) was highes in RDCFIN (15.5%),
ollowed by RDCSWE (10.1 %) and RDCDNK (3.9%).
No homozygous indi iduals o his haplo ype we e
obse ed. When his haplo ype was included in he model
as a co ac o , no addi ional signi ican associa ions we e
Table 2 Top 10 mos signi ican SNPs selec ed ocons uc
haplo ypes o haplo ype-based associa ion mapping
Resul s a e o cal su i al a la e lac a ions
E ec s and SE a e gi en in b eeding alue index uni s
MAF mino allele equency, SE s anda d e o
SNP MAF Allele subs i u ion
e ec SE p alue
Ch 23:13,259,463 0.062 −10.15 0.43 4.11e−115
Ch 23:13,260,528 0.062 −10.15 0.43 4.05e−115
Ch 23:13,260,771 0.062 −10.15 0.43 4.02e−115
Ch 23:13,261,303 0.062 −10.15 0.43 4.03e−115
Ch 23:13,264,527 0.062 −10.15 0.43 4.14e−115
Ch 23:13,269,718 0.062 −10.15 0.43 3.48e−115
Ch 23:13,270,817 0.068 −10.22 0.44 2.39e−115
Ch 23:13,271,582 0.062 −10.15 0.43 3.48e−115
Ch 23:13,313,896 0.059 −10.56 0.44 1.00e−120
Ch 23:13,354,932 0.066 −9.84 0.42 3.30e−113
Fig. 3 Single‑ a ian associa ion signal plo s o cal su i al a la e
cal ings (SBL) (ligh g ay do s) and single‑ a ian analysis wi h he
10‑SNP haplo ype as a co ac o (blue do s). Haplo ypes we e con‑
s uc ed wi h he 10 op associa ed SNPs a he QTL peak and i ed
as co a ia es in he model (blue do s)
Fig. 4 Single‑ a ian associa ion signal plo s o bi h index (BI) (ligh
g ay do s) and single‑ a ian analysis wi h he 10‑SNP haplo ype as
a co ac o (blue do s). Haplo ypes we e cons uc ed wi h he 10 op
associa ed SNPs a he QTL peak and i ed as co a ia es in he model
(blue do s)
Fig. 5 Single‑ a ian associa ion signal plo s o cal su i al a i s
cal ing (SBF) (ligh g ay do s) and single‑ a ian analysis wi h he
10‑SNP haplo ype as a co ac o (blue do s). Haplo ypes we e con‑
s uc ed wi h he 10 op associa ed SNPs a he QTL peak and i ed
as co a ia es in he model (blue do s)
Page 7 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
obse ed (blue do s in Figs.3, 4, 5), sugges ing ha his
haplo ype could ully explain he QTL a iance.
Emb yonic homozygosi y o HAPQTL is esponsible
o s illbi h
We analyzed he s illbi h a e o he ou ma ing ypes
(I, II, III and IV) based on he ca ie s a us o he si e
and ma e nal g andsi e. The s illbi h a e was app oxi-
ma ely 6% highe o ma ings in which bo h he si e and
ma e nal g andsi e ca ied he pu a i e causa i e hap-
lo ype HAPQTL compa ed o ma ings in which nei he
was a ca ie (Fig.6). We consis en ly obse ed highe
cal mo ali y a es om ca ie -si e by ca ie -ma e nal
g andsi e ma ings ac oss he h ee No dic coun ies, in
bo h he i s and la e cal ings (Table3).
P esence o a la ge ch omosomal dele ion in he QTL
egion
De ia ion omHWP
Signi ican de ia ions om HWP we e obse ed o
SNPs on he HD chip in he in e al be ween 12.1 and
12.9 Mbp (see Addi ional ile3: Figu e S3), all o which
we e due o an excess o homozygo es o he SNPs. This
esul p o ides a s ong indica ion o a ch omosomal
dele ion.
HD geno ype in ensi y con i ms he s uc u al a ia ion
The plo o he a e age in ensi ies om ca ie s and
nonca ie s o CHRDEL (Fig.7) showed a clea educ-
ion in signal in ensi ies o SNPs in he egion be ween
12,289,281 and 12,816,660 bp, which con i med he
exis ence o a ch omosomal dele ion. Since he dele ed
egion ha bo ed se e al epe i i e elemen s, i was di -
icul o iden i y he exac b eaking poin s om he
posi ion o he spli eads o he sequence da a. Ampli-
ica ion ac oss he dele ed egion was unsuccess ul
because unique p ime s o i s ex emi ies could no be
designed. Sequence co e age inc eased in dep h a he
end egion o he b eaking poin (see Addi ional ile4:
Figu e S4), which could be due o assembly p oblems o
he p esence o epe i i e elemen s. These da a, com-
bined wi h he indi idual HD SNP in ensi y da a and
sequence alignmen s ( esul s no shown), indica e ha
ei he he bo ine genome assembly o his egion is
inco ec , o ha he s uc u al a ia ion is mo e com-
plex han jus a simple dele ion.
The dele ion disco e ed he e (CHRDEL) is no he
only s uc u al a ia ion ound in he egion. Bickha
e al. [33] published a copy numbe a ia ion (CNV) in
Angus ca le ha o igina ed om G ea B i ain wi hin
he egion a ec ed by CHRDEL (BTA23: 12,500,344–
12,515,633bp), and Hou e al. [34] ound a CNV (BTA23:
12,938,090–13,081,504 bp) in No dic Red dai y ca le
nea CHRDEL. Bo h CNV a e gains o DNA sequence and
a e much smalle han he s uc u al a ia ion iden i ied
he e.
In addi ion, we checked he geno ype in ensi y (log2R
a io) o eigh SNPs on he 50k chip loca ed wi hin he
dele ed egion in a la ge sample o 13,852 No dic Red
dai y ca le animals. The dis ibu ion o he in ensi ies
is in Fig.8. Animals ha we e homozygous o all eigh
SNPs had low geno ype in ensi ies, which suppo s he
Fig. 6 S illbi h a e (%) by ma ing ype acco ding o he ca ie s a us o he si e (S) and ma e nal g andsi e (MGS) o he le hal haplo ype (HAPQTL),
based on he op 10 SNPs associa ed wi h s illbi h. I nonca ie S × MGS, II nonca ie S × ca ie MGS, III ca ie S × nonca ie MGS and IV ca ie
S × MGS ma ings, FIN Finland, SWE Sweden and DNK Denma k
Page 8 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
exis ence o a ch omosomal dele ion. The leng h o his
dele ion was app oxima ely 0.5Mbp, as indica ed by he
HD in ensi y da a and WGS dep h.
50k geno ype da a con i m he p esence o a
ecessi e le hal haplo ype spanning he dele ed
egion
We iden i ied a haplo ype (HAP50k) ha spanned a
egion be ween 11,818,761 and 13,122,318 bp wi h
a popula ion equency o 0.043, bu we ound no
homozygous indi iduals o his haplo ype. Unde
Ha dy–Weinbe g equilib ium, we would expec o ha e
36 homozygo es o his haplo ype among he 19,309
50 k-geno yped animals. The p obabili y o obse ing
no homozygo es by chance is 2.31×10−16 (assuming
a Poisson dis ibu ion), i his haplo ype is no a eces-
si e le hal a ian . Absence o homozygo es o HAP50k
is s ong e idence ha Hap50k ei he ca ies o ags a
ecessi e le hal a ian .
Conco dance o ca ie s o HAPQTL andCHRDEL
WGS dep h con i ms he ch omosomal dele ion
We ook ad an age o he WGS in o ma ion ha was
a ailable o 83 No dic Red dai y bulls. Six sequenced
bulls ca ied he haplo ype associa ed wi h s illbi h
(HAPQTL). All six also showed an app oxima ely hal ed
sequencing dep h be ween 12.29 and 12.84 Mbp (one
example shown in Addi ional ile4: Figu e S4), whe eas
such a educ ion was no ound o any o he nonca ie
bulls. None o he 59 Hols ein o Je sey bulls o which
WGS da a was analyzed, exhibi ed his educ ion in
sequencing dep h o his egion on BTA23.
CHRDEL ca ie s in e ed omHD geno ype in ensi y
Among he 243 RDCFIN bulls wi h HD geno ypes, 47
animals we e decla ed ca ie s based on he c i e ia
o an a e age geno ype signal in ensi y less han 0 and
homozygosi y o mo e han 134 o he 139 loci wi hin
he pu a i e dele ed egion be ween 12.2 and 12.8Mb. O
Table 3 Numbe o obse a ions (N) andcal su i al a es (Su i al) (%) o ma ing ypes acco ding ogeno ypes o  he
si e (S) andma e nal g andsi e (MGS) o  he 500-kb dele ion
C ca ie , NC non ca ie , S si e, MGS ma e nal g and si e, FIN Finland, SWE Sweden, DNK Denma k
Mo he Ma ing ype To al
S‑NC×MGS‑NC (I) S‑NC×MGS‑C (II) S‑C×MGS‑NC (III) S‑C×MGS‑C (IV)
N Su i al N Su i al N Su i al N Su i al
DNK‑hei e 57,165 94.3 1971 94.9 2014 94.6 73 93.2 61,223
DNK‑cow 84,679 96.9 2551 96.7 3904 96.9 176 94.3 91,310
FIN‑hei e 535,406 94.1 120,843 93.6 105,221 93.0 21,459 88.3 782,929
FIN‑cow 805,528 96.3 178,770 96.0 178,395 95.3 36,778 90.4 1,199,471
SWE‑hei e 496,779 95.8 53,709 95.3 44,785 95.7 3250 90.7 598,523
SWE‑cow 828,052 96.7 97,536 96.5 81,253 96.4 8373 91.9 1,199,471
To al 2,785,085 95.9 536,614 95.6 512,714 95.3 98,514 90.5 3,932,927
Fig. 7 A e age geno ype signal in ensi y pe SNP o ca ie s ( ed) and nonca ie s (blue) o he ch omosomal dele ion (CHRDEL), plo ed agains
hei genomic posi ions. Ca ie s a us o an indi idual was p edic ed based on loss o geno ype in ensi y and homozygosi y using SNPs om he
Illumina Bo ineHD geno yping a ay
Page 9 o 12
Sahana e al. Gene Sel E ol (2016) 48:35
hese 47 ca ie s, 25 we e homozygous o all 139 loci, 13
we e homozygous o 138 loci, and nine we e homozy-
gous o 134–137 loci. We examined he conco dance
be ween ca ie s o CHRDEL and ca ie s o he ecessi e
le hal haplo ype based on 50k da a (HAP50k). Among he
47 animals ha ca ied he dele ion, 42 ca ied he ag-
ging haplo ype. SNPs om he Illumina Bo ineSNP50
BeadChip loca ed wi hin he dele ed egion a e in Addi-
ional ile5: Table S2.
Discussion
A la ge-e ec QTL o BI was de ec ed on BTA23 in a
genome-wide associa ion s udy in No dic Red dai y ca -
le. A de ailed s udy o i s componen ai s showed ha i
had a la ge e ec on s illbi h. We cons uc ed haplo ypes
om he 10 SNPs ha we e mos s ongly associa ed
wi h s illbi h based on WGS da a and iden i ied animals
ha ca ied he haplo ype ha had a nega i e e ec on
s illbi h (HAPQTL). No homozygo es o HAPQTL we e
obse ed. No o he QTL e ec s we e obse ed in he
egion when HAPQTL was included in he model as a
co ac o . Ma ings o which he si e and ma e nal g and-
si e ca ied HAPQTL had signi ican ly highe equencies
o s illbi h compa ed o ma ings o which he si e and/
o ma e nal g andsi e we e a nonca ie .
We used he HD SNP geno ypes o sea ch o ch omo-
somal dele ions wi hin he QTL egion. Based on de ia-
ions om HWP and educed geno ype in ensi y, we
obse ed a la ge ch omosomal dele ion (CHRDEL) and
iden i ied animals ha ca ied his dele ion. Using a la ge
numbe o animals wi h 50k geno ypes, we con i med
he exis ence o a ecessi e le hal haplo ype (HAP50k)
ha spanned he dele ed egion and iden i ied ca ie s o
HAP50k.
A e s udying he conco dance be ween animals ca -
ying HAPQTL, CHRDEL and HAP50k, an app oxima ely
500-kb dele ion on BTA23 eme ged as a s ong candi-
da e QTL o s illbi h. Two lines o e idence suppo his
conclusion. Fi s , among he 83 sequenced No dic Red
dai y ca le animals, he six bulls ha ca ied HAPQTL
showed a window whe e ead dep h was educed by 50%
compa ed o he es o he genome. In con as , no non-
ca ie bulls showed his educ ion in sequencing dep h.
Howe e , unce ain y emains on he exac posi ion o
he b eakpoin s o he dele ed egion due o he p es-
ence o o he s uc u al a ian s in his genomic egion
(see Addi ional ile4: Figu e S4). Second, analysis o 50k
geno ypes o a la ge numbe o indi iduals con i med
a ecessi e le hal haplo ype (HAP50k) ha spanned his
dele ed egion.
Se e al obse a ions suppo he conclusion ha
HAPQTL is a ecessi e le hal haplo ype ha causes s ill-
bi h. Fi s , no homozygo es o HAPQTL we e obse ed.
Second, he e ec o HAPQTL was obse ed in he case
o ca ie -si e and ca ie -ma e nal g andsi e ma ings o
all h ee No dic Red dai y ca le popula ions. The e was
a sligh inc ease in s illbi hs among ype III ma ings in
he Finnish popula ion. This esul may be due o a highe
equency o ca ie s o HAPQTL, such ha he dam
inhe i s he dele ion om i s dam (i.e. ma e nal g and-
dam o he cal ) e en when he ma e nal g andsi e is a
nonca ie . F equencies o ca ie s o HAPQTL we e much
lowe in he Swedish and Danish RDC popula ions. Con-
sequen ly, no e ec on e ili y was obse ed in ype III
ma ings in hese popula ions.
The inc ease in he a e o s illbi hs among ype IV
ma ings was smalle han he 12.5% expec ed o a ully
pene an ecessi e le hal allele. Because s illbi hs a e
eco ded by a me s, he di e ence may be due o some
o he s illbi hs no being epo ed o he cen al eg-
is a ion sys em. I is also possible ha some dea hs o
cal es occu ed du ing la e ges a ion and, hus, we e
no eco ded by a me s as s illbi hs. The s illbi h is
eco ded as a bina y ai bu we analyzed i by assum-
ing a no mal dis ibu ion o es he e ec o ma ing
ype on cal su i al a bi h. A logis ic eg ession model
migh ha e been a be e choice o he analysis o his
pheno ype.
Genes loca ed wi hin he dele ed egion
Th ee genes a e loca ed wi hin he dele ed egion:
(1) BTB (POZ) domain con aining 9 (BTBD9,
Fig. 8 Densi y plo o geno ype in ensi y o eigh SNPs loca ed
wi hin he ch omosomal dele ion in he Illumina Bo ineSNP50 Bead‑
Chip o an independen sample o No dic Red dai y ca le