Seasonal in luenza ci cula ion pa e ns and p ojec ions
o Sep embe 2025 o Sep embe 2026
John Huddles on
1
, Jenni e Chang
1
, Jo e Lee
1
, Philippa S einbe g
1
, Richa d A. Nehe
2
& T e o Bed o d1,3
1Vaccine and In ec ious Disease Di ision, F ed Hu chinson Cance Cen e , Sea le, WA, USA, 2Biozen um,
Uni e si y o Basel, Basel, Swi ze land, 3Howa d Hughes Medical Ins i u e, Sea le, WA, USA
Sep embe 23, 2025
Abs ac
This epo de ails cu en seasonal in luenza ci cula ion pa e ns as o mid-Sep embe 2025 and
was p epa ed o he VCM ahead o he Sou h Hemisphe e VCM on Sep embe 19, 2025. This is no
mean as a comp ehensi e epo , bu is ins ead in ended as pa icula obse a ions ha we’ e made
ha may be o ele ance. Please also no e ha obse ed pa e ns e lec he GISAID da abase and
may no be en i ely ep esen a i e o unde lying dynamics. All analyses a e based on he Nex s ain
pipeline [1,2] wi h con inual upda es pos ed o nex s ain.o g/seasonal- lu. In compliance wi h da a
sha ing ag eemen s, his public e sion o he epo does no include aw se ological measu emen s.
A/H1N1pdm: D.3 con inues o sweep globally likely due o easso men wi h di e en NA
backg ounds. We de ine he new HA clade D.3.1 o ack his easso an g oup. This clade
con inues o di e si y gene ically wi h a leas one majo subclade ising o high equency in Eu ope,
ca ying 113K, 139D, and 283K subs i u ions. A/Wisconsin/67/2022 e ec i ely co e s ecen i uses
excep hose ca ying HA1:155E o 155R subs i u ions. Ou p edic ion models sugges ha accine
candida es in he D.3.1 subclade bea ing 113K ep esen he closes an igenic ma ch o he u u e
popula ion. A/H3N2: J.2.4 (wi h HA1 189R and 135K) is g owing apidly in A ica, Eu ope,
No h Ame ica, and Oceania, shows s ong an igenic d i in e e s, and is p edic ed o ix in hese
egions. J.2 wi h 145N and 261Q is g owing apidly in Sou heas Asia and shows s ong an igenic
d i in humans. J.2.4, J.2.3 (189R and 158K), and J.2.5 (145N and 158K) a e poo ly co e ed by
A/Dis ic O Columbia/27/2023 and A/C oa ia/10136RV/2023. We p edic J.2.4 accine candida es
like A/Sydney/1359/2024, A/Nepal/N042/2025-egg, and A/Singapo e/GP20238/2024-egg a e bes
ma ched o he u u e popula ion, al hough se a o hese s ains poo ly co e o he ci cula ing
i uses. B/Vic: The subclade C.3.1 is g owing in No h Ame ica wi h a s ong signal o an igenic
d i om e e HI da a associa ed wi h an HA1 197N subs i u ion. The newly de ined subclade
C.3.1 shows e idence o easso men o he NA backg ound o HA subclade C.5.1. C.5.7 and C.5.1
appea o be declining globally as C.5.6.1 g ows. Se a agains B/Pennsyl ania/14/2025 appea s o
co e C.3.1 and o he ex an clades.
Con en s
Me hods .......................................... 3
A/H1N1pdm ....................................... 7
Cu en ci cula ion pa e ns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
An igenic p ope ies om e e and indi idual human se ology . . . . . . . . . . 10
Compa ing i al i ness o ecas s and an igenic pheno ype . . . . . . . . . . . . . 12
1
A/H3N2 .......................................... 14
Cu en ci cula ion pa e ns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
An igenic p ope ies om e e and indi idual human se ology . . . . . . . . . . 17
Compa ing i al i ness o ecas s and an igenic pheno ype . . . . . . . . . . . . . 18
B/Vic ............................................ 21
Cu en ci cula ion pa e ns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
An igenic p ope ies om e e and indi idual human se ology . . . . . . . . . . 24
Compa ing i al i ness o ecas s and an igenic pheno ype . . . . . . . . . . . . . 26
2
Me hods and No es
Sequence da a and subsampling
We base ou analysis on sequence da a a ailable in GISAID as o Sep embe 18, 2025 and i e da a
a ailable as o Sep embe 18. The a ailabili y o sequences a ies g ea ly ac oss ime and geog aphy
and we y o minimize geog aphical and empo al bias by subsampling he da a o analyzing di e en
geog aphical egions sepa a ely when app op ia e. While his subsampling educes geog aphical biases,
i doesn’ emo e his bias en i ely.
Phylogene ic analysis
The da abase con ains oo many sequences o pe o m a comp ehensi e phylogene ic analysis o all
a ailable da a. We hence subsample he da a o a leas 3000 sequences collec ed in he las 2 yea s.
To minimize bias om di e ences in su eillance in ensi y, we e enly sample a mos 2700 sequences
(90%) ac oss egion, yea and mon h bins o egions including No h Ame ica, Oceania, China, Sou h
Ame ica, Wes Asia, Japan and Ko ea, Sou heas Asia, A ica, and Sou h Asia. Fo Eu opean da a, we
sample a mos 300 sequences (10%) ac oss coun y, yea and mon h bins. Wi hin each egion/coun y,
yea , and mon h g oup, we p io i ize sequences wi h co esponding an igenic da a. Fo phylogene ic
con ex , we include an addi ional 300 sequences collec ed p io o 2 yea s ago and as ea ly as Janua y
2016, e enly sampling by egion and yea . We also include all s ains wi h i e measu emen s as a es
o e e ence i us, o ensu e ha all i e da a a e a ailable o obse e in ou in e ac i e epo s.
Pa allel e olu ion, ha is epea ed occu ence o iden ical subs i u ions in di e en clades o he ee,
is common in A/H3N2 and A/H1N1pdm. Such pa allel e olu ion iola es undamen al assump ions o
common phylogeny so wa e and can e oneously g oup dis inc clades oge he i hey sha e oo many
pa allel changes. To a oid such a i ac s, we mask si es wi h ampan pa allelism p io o phylogeny
in e ence.
Clades and subclades a e assigned using a collec ion o “signa u e” mu a ions a ailable a he Gi Hub
eposi o y gi hub.com/nex s ain/seasonal- lu o each lineage ia he ollowing links:
•H1N1pdm HA
•H1N1pdm NA
•H3N2 HA
•H3N2 NA
•Vic HA
•Vic NA
F equency and i ness es ima es
Al hough phylogene ic analyses equi e a subse o all a ailable sequences o p oduce an unbiased
ee, we apply a non-phylogene ic me hod o es ima e clade equencies om all hemagglu inin
(HA) sequences pe lineage. We align all HA sequences o he lineage’s e e ence sequence using
Nex clade and ob ain clade labels and amino acid subs i u ions pe sequence [3]. F om hese Nex clade
anno a ions pe sequence, we join ly es ima e he equencies and ela i e g ow h ad an ages o each
clade pe geog aphic egion in 14-day in e als using a hie a chical mul inomial logis ic eg ession
3
(MLR) model [4]. Mul inomial logis ic eg ession ac oss
n
a ian s models he p obabili y o a i us
sampled a ime belonging o a ian ias equal o i s equency xi( ) ollowing
P (X=i) = xi( ) = piexp( i )
Pjpjexp( j ).
A oy example o MLR i o equency da a is shown in Figu e 1.
Figu e 1. Demo o MLR model and di ec ly es ima ing clade-le el g ow h ad an age om
equency dynamics.
G ow h ad an ages e lec he ad an age o each clade ela i e o a speci ic e e ence clade o “pi o ”
clade. Values less han 1 indica e lowe i ness han he pi o clade and alues g ea e han 1 indica e
highe i ness. In addi ion o es ima ing egion-speci ic g ow h ad an ages, he model es ima es an
a e age g ow h ad an age and a iance pe clade ac oss all egions which we epo as he “hie a chical
GA”. The hie a chical GA is analogous o he local b anching index in ha i e lec s he ecen global
success o each clade. The egion-speci ic GA is analogous o a egion-speci ic local b anching index.
An igenic analysis o e e da a
We summa ize HI and FRA measu emen s p o ided by he WHO CCs in London, Melbou ne, A lan a,
and Tokyo using ou subs i u ion model [5] which models log- i e s as a sum o e ec s associa ed
wi h amino acid di e ences be ween he sequences o he es and e e ence i us. In addi ion, he
model allows o a se um (column) and a i us ( ow) e ec . This model allows in e ence o i e s
o i us/se um pai s ha ha e no been an igenically cha ac e ized and isola es e ec s consis en ly
obse ed ac oss many measu emen s om he noise inhe en in indi idual measu emen s. We also plo
indi idual and a e age no malized
log2
i e s pe e e ence se um o es i uses in speci ic clades, o
ep esen he aw an igenic da a a ailable o each in luenza lineage.
In addi ion o he MLR-based g ow h ad an ages desc ibed abo e, we es ima ed he i ness o cu en ly
ci cula ing s ains using an igenic ad ance based on i e measu emen s om e e se a [5]. High
an igenic ad ance indica es g oups o i uses wi h amino acid subs i u ions ha a e in e ed o inc ease
he an igenic dis ance om i uses wi hou hose subs i u ions.
4
High h oughpu human se ology and an igenic analysis
Kikawa e al. [6] desc ibes he gene a ion o a lib a y o 76 ecen ly ci cula ing H3N2 s ains and
38 ecen ly ci cula ing human H1N1pdm s ains, as well as 26 his o ical accine s ains o H3N2
and H1N1pdm (Fig. 2A). This wo k uses a high- h oughpu sequencing-based neu aliza ion assay o
measu e neu aliza ion i e s o hese 140 i uses agains a panel o 188 human se a collec ed om
indi iduals spanning om young child en o elde ly adul s, d awn om ou si es a ound he wo ld
and collec ed be ween Oc obe 2024 and Ap il 2025. Va ia ion in i e is clea ac oss indi iduals and
ac oss i uses (Fig. 2B). We apply he same ‘ i e model’ as used o e e da a [5] o p edic an igenic
escape o s ains ha we e no di ec ly measu ed in he assay. In his epo we ocus on a subse o
es i uses which a e cell-passaged (N=39 H1N1pdm and N=77 H3N2) and he subse o human se a
ha we e no known o be pos - accina ion (N=181).
Figu e 2. High- h oughpu sequencing-based neu aliza ion o measu e human se ological e-
sponses ac oss i uses. (A) Sequencing-based neu aliza ion assays enable apid measu emen o i e s agains
many s ains. (B) Ti e s o indi idual human se a o ecen H3N2 s ains. Figu e ep oduced om Kikawa e
al. [6].
5
An igenic dis ance o he p edic ed u u e popula ion
To ank accine candida es pe sub ype by hei an igenic dis ance o he u u e, we in eg a ed
an igenic dis ance es ima ed om human o e e se ological da a wi h p edic ed u u e equencies
es ima ed by he MLR model. The MLR model allows us o p edic u u e equencies o eme ging
haplo ypes pe geog aphic egion. Ou i e subs i u ion model [5] es ima es he an igenic e ec o each
HA1 subs i u ion be ween e e ence and es i uses. Fo each e e ence i us used in e e cell HI
expe imen s ( accine candida es), we calcula ed he a e age an igenic dis ance be ween ha candida e’s
HA1 amino acid sequence and he co esponding HA1 sequence o each i us in each combina ion o
eme ging haplo ype and geog aphic egion. We calcula ed each pai wise an igenic dis ance be ween he
candida e and ano he i us sequence as he sum o an igenic e ec s associa ed wi h he subs i u ions
be ween he wo i al HA1 sequences. We calcula ed a weigh ed a e age an igenic dis ance o he u u e
by mul iplying he p edic ed u u e equency o each haplo ype in a gi en egion (e.g., Fig. 4) by he
a e age an igenic dis ance o ha haplo ype/ egion combina ion. Finally, we a e aged hese es ima es
ac oss egions o ge he weigh ed a e age an igenic dis ance o he accine s ain o he p edic ed
u u e popula ion. Candida es wi h he lowes dis ances a e p edic ed o be he bes ep esen a i es o
he u u e popula ion.
6
A/H1N1pdm
D.3 con inues o sweep globally likely due o easso men wi h di e en NA backg ounds.
We de ine he new HA clade D.3.1 o ack his easso an g oup. This clade con inues o
di e si y gene ically wi h a leas one majo subclade ising o high equency in Eu ope,
ca ying 113K, 139D, and 283K subs i u ions. A/Wisconsin/67/2022 e ec i ely co e s
ecen i uses excep hose ca ying HA1:155E o 155R subs i u ions. Ou p edic ion
models sugges ha accine candida es in he D.3.1 subclade bea ing 113K ep esen he
closes an igenic ma ch o he u u e popula ion.
Cu en ci cula ion pa e ns
In he six mon hs since he las VCM, D.3 has swep o ixa ion in e e y egion excep A ica, China,
and Sou heas Asia whe e i is g owing apidly (Figs. 3 and 4). The ecen ly success ul subclade D.3.1
lacks HA1 subs i u ions ela i e o D.3 ha could explain i s success, bu his clade has easso ed wi h
mul iple NA backg ounds including he NA subclade D.1 ( o me ly C.5.3.1 wi h A98V subs i u ion).
These easso men e en s may explain D.3.1’s consis en ly highe g ow h ad an age o e all o he
clades ac oss all egions (Fig. 5).
Now ha D.3 has essen ially ixed, we no e h ee eme ging haplo ypes including D.3.1:113K, D.3.1:113K-
139D-293K, and D.3.1:185T. D.3.1:113K has easso ed wi h wo di e en NA backg ounds including a
small subclade wi h NA:V263I and he la ge NA subclade D.2 (D wi h NA:52N). This clade ne e
eached high equency globally his yea and con inues o ha e a lowe g ow h ad an age ac oss all
egions (Fig. 5). In con as , D.3.1:113K-139D-293K has ci cula ed in i e egions including Sou heas
Asia, has easso ed wi h an NA subclade D wi h a M314I subs i u ion, and has he highes g ow h
ad an age ac oss all egions (Fig. 5). A subclade o D.3.1:113K-139D-293K ca ies an addi ional
HA1:K302E subs i u ion ha is associa ed p ima ily wi h ecen Eu opean sequences. D.3.1:185T
has only been collec ed in Aus alia and New Zealand and ci cula es wi h an NA subclade D.1 wi h
A81V. Gi en i s limi ed geog aphic dis ibu ion, we can only say ha D.3.1:185T’s g ow h ad an age
is simila o D.3.1 in Oceania. The HA1 si es 139 and 185 a e bo h pu a i e epi ope si es.
A subse o ecen D.3.1 i uses ha e acqui ed a HA1:157L subs i u ion. These i uses ha e been
collec ed p ima ily in Denma k (N=67) wi h ewe han 5 samples collec ed in he USA, Aus alia,
Thailand, o he Philippines. As a esul o he ecen samples om Denma k, we es ima e a high
g ow h ad an age o D.3.1:157L i uses (Fig. 5). We also obse e modes an igenic ad ance o hese
i uses om i e models i o se ological da a om bo h e e s (Fig. 6) and humans (Fig. 7). Gi en
he global a i y o his subs i u ion a he momen , we do no ha e enough e idence o be su e o i s
i ness compa ed o o he haplo ypes.
7
Figu e 3. Time- esol ed A/H1N1pdm phylogeny colo ed by clade and il e ed o s ains collec ed
since Feb ua y 1, 2025 ( op) and co esponding geog aphic dis ibu ion o s ains shown in he
phylogeny (bo om). View on nex s ain.o g.
8
Figu e 4. Clade equencies by egion es ima ed by a mul inomial logis ic eg ession (MLR) model.
Lines show he median equency es ima ed unde an assump ion o exponen ial g ow h. Shading indica es 95%
lowe and uppe highes pos e io densi y in e als om he model. Colo ed poin s indica e equency es ima es
om he aw da a binned in o biweekly in e als. Selec ion mani es s as s aigh lines on he logi scale o he
y-axis whe e he slope co esponds o he s eng h o selec ion pe clade. Fo ecas s show MLR model p edic ions
a biweekly s eps up o 1 yea in he u u e. View on Nex s ain.
Figu e 5. G ow h ad an ages pe clade by egion as es ima ed by a MLR model. The “hie a chical”
ad an age ep esen s he global a e age i ness pe clade. The dashed e ical line a x=1 indica es he g ow h
ad an age o he pi o clade. Clades wi h g ow h ad an ages g ea e han 1 ha e a highe i ness han he pi o .
Colo ed poin s indica e he median g ow h ad an age pe clade and loca ion. E o ba s indica e he 95% lowe
and uppe highes pos e io densi y in e als om he model. View on Nex s ain.
9
Figu e 12. Clade equencies by egion es ima ed by mul inomial logis ic eg ession (MLR). Lines
show he median equency es ima ed unde an assump ion o exponen ial g ow h. Shading indica es 95% lowe
and uppe highes pos e io densi y in e als om he model. Colo ed poin s indica e equency es ima es om
he aw da a binned in o biweekly in e als. Selec ion mani es s as s aigh lines on he logi scale o he y-axis
whe e a clade’s slope co esponds o i s s eng h o selec ion. Fo ecas s show MLR p ojec ions up o 1 yea in
he u u e. View on Nex s ain.
Figu e 13. G ow h ad an ages pe clade by egion as es ima ed by MLR. The “hie a chical” ad an age
ep esen s he global a e age i ness pe clade. The dashed e ical line a x=1 indica es he g ow h ad an age
o he pi o clade. Clades wi h g ow h ad an ages g ea e han 1 ha e a highe i ness han he pi o . Colo ed
poin s indica e he median g ow h ad an age pe clade and loca ion. E o ba s indica e he 95% lowe and
uppe highes pos e io densi y in e als om he model. View on Nex s ain.
16
An igenic p ope ies om e e and indi idual human se ology
[Raw e e da a edac ed]
When we calcula ed an igenic ad ance, agg ega ing e e -based cell-passaged HI da a om all collabo-
a ing cen e s, we ound he highes ad ance associa ed wi h s ains ca ying 158K and 189R including
subclades J.2.3, J.2.4, and J.2.5 (Fig. 14). Mos i uses om he J.2.4:158D-160K subclade ca y he
ollowing i e HA1 subs i u ions wi h nonze o an igenic e ec s om he i e subs i u ion model. These
subs i u ions a e lis ed below wi h hei log2e ec s:
•K158D:1.43
•I160K: 0.01
•T328A: 0.72
•S144N: 0.76
•Q173R: 0.22
We also i i e models o neu aliza ion i e s o ecen ly collec ed human se a and H3N2 i uses
om Kikawa e al. [6]. Unlike he e e da a, human models show lowe an igenic ad ance o J.2.4
co esponding o lowe es ima ed e ec s o 135K and 189R (Fig. 15). The i us lib a y o hese
neu aliza ion expe imen s lacked i uses om J.2.4:158D-160K, so we canno commen on he an igenic
ad ance o he subs i u ions abo e om hose da a. The human models also show a sligh ly highe
an igenic ad ance o J.2:145N-261Q i uses which could pa ially explain he ecen success o hese
i uses in Sou heas Asia.
Figu e 14. An igenic ad ance pe clade es ima ed om e e -based HI da a [5] o samples
collec ed since Feb ua y 1, 2025. In each panel, poin s ep esen indi idual HA sequences whose o e all
an igenic ad ance has been in e ed om a i e model based on he co esponding aw se ological da a.
17
Figu e 15. An igenic ad ance pe clade es ima ed wi h a i e subs i u ion model [5] om human-
based neu aliza ion da a [6].
Compa ing i al i ness o ecas s and an igenic pheno ype
To unde s and he ela ionship be ween i al i ness es ima ed om sequence da a alone (g ow h
ad an ages om he MLR model) o om expe imen al da a (an igenic ad ance), we plo ed he g ow h
ad an age o each ecen eme ging haplo ype agains he co esponding median an igenic ad ance
o sequences in ha haplo ype based on human o e e se ological da a. Wi h bo h measu es o
an igenic ad ance, we ind ha g ow h ad an ages om sequence da a gene ally co espond well o
an igenic ad ance (Fig. 16). J.2.5 has one o he highes an igenic ad ances by bo h species models,
bu i also has one o he lowes g ow h ad an ages. The human da a / model inds highe an igenic
ad ance o J.2:145N-261Q han he e e model and ails o iden i y signi ican an igenic ad ancemen
o J.2.4 i uses despi e hei apid g ow h. The e e model iden i ies highe an igenic ad ance o
bo h J.2.4 and J.2.4:158D-160K ha co espond di ec ly o he g ow h ad an ages we es ima e o
hese haplo ypes. Howe e , he e e model also iden i ies highe an igenic ad ance o J.2.3 han he
human model despi e he lack o g ow h o his clade.
Finally, we anked he a ailable accine candida es by hei an igenic dis ance o he p edic ed u u e
H3N2 popula ion as desc ibed in he Me hods. Resul s om bo h human and e e an igenic models,
ound ha J.2.4 e e ence i uses like A/Sydney/1359/2024 we e closes o he p edic ed u u e
(Figs. 17 and 18). Fe e -based an igenic dis ances also ound ha ea lie J.2 i uses wi h he K189R
subs i u ion like A/Idaho/69/2023 we e close o he u u e popula ion. This esul likely e lec s he
ac ha e e da a de ec highe an igenic ad ance o subclades wi h 189R like J.2.3 and J.2.4 han
he human da a de ec .
18
Figu e 16. G ow h ad an age by an igenic ad ance pe clade es ima ed wi h a i e subs i u ion
model [5] o human-based neu aliza ion da a [6] (le ) and e e -based cell HI da a ( igh ).
G ow h ad an ages shown as median +/- 95% HPDIs. An igenic ad ance shown as median +/- 1 s anda d
de ia ion ac oss all sequences in a gi en clade.
Figu e 17. Weigh ed a e age an igenic dis ance o he u u e popula ion pe e e ence i us using
human-based neu aliza ion da a. Fo each e e ence i us, we calcula ed he a e age an igenic dis ance
be ween ha i us’s HA1 amino acid sequence and he co esponding HA1 sequences o i uses om each
eme ging haplo ype and geog aphic egion. The pai wise an igenic dis ance be ween i uses was based on he
HA1 subs i u ions be ween hem and an igenic weigh s pe subs i u ion es ima ed wi h a i e subs i u ion
model [5] i o human-based neu aliza ion da a [6].
19
Figu e 18. Weigh ed a e age an igenic dis ance o he u u e popula ion pe e e ence i us
using e e -based cell HI da a. Fo each e e ence i us, we calcula ed he a e age an igenic dis ance
be ween ha i us’s HA1 amino acid sequence and he co esponding HA1 sequences o i uses om each
eme ging haplo ype and geog aphic egion. The pai wise an igenic dis ance be ween i uses was based on he
HA1 subs i u ions be ween hem and an igenic weigh s pe subs i u ion es ima ed wi h a i e subs i u ion
model [5] i o e e -based cell HI da a.
20
B/Vic
The subclade C.3.1 is g owing in No h Ame ica wi h a s ong signal o an igenic d i
om e e HI da a associa ed wi h an HA1 197N subs i u ion. The newly de ined
subclade C.3.1 shows e idence o easso men o he NA backg ound o HA subclade
C.5.1. C.5.7 and C.5.1 appea o be declining globally as C.5.6.1 g ows. Se a agains
B/Pennsyl ania/14/2025 appea s o co e C.3.1 and o he ex an clades.
Cu en ci cula ion pa e ns
Al hough clades C.5.1, C.5.6, and C.5.7 con inue o coci cula e globally, clade C.3.1 has eeme ged
and g own in No h Ame ica (Figs. 19 and 20). In his epo , we de ine h ee new subclades o ack
a ian s wi h signals o an igenic d i om e e se ology:
•C.5.6.1 wi h HA1 37I, 128D, and 199A
•C.3.1 wi h HA1 197N and 208S
•C.3.2 wi h HA1 197N, 179H, and 208P
Mos ecen C.3 i uses descend om C.3.1 and appea o ha e easso ed om C.3’s ances al NA
subclade o B.8 o C.5.1’s NA subclade B.7.1. The ecen g ow h o C.3 and C.3.1 con ibu es hei
high es ima ed i nesses ela i e o he o he h ee majo clades (Fig. 21). C.3.1 does no appea o
ha e sp ead widely ou side o he USA.
21
Figu e 19. Time- esol ed B/Vic o ia phylogeny colo ed by clade and il e ed o s ains collec ed
since Feb ua y 1, 2025 ( op) and co esponding coun y-le el geog aphic dis ibu ion o s ains
shown in he phylogeny (bo om). View on nex s ain.o g.
22
Figu e 20. Clade equencies by egion es ima ed by a mul inomial logis ic eg ession (MLR)
model. Lines show he median equency es ima ed unde an assump ion o exponen ial g ow h. Shading
indica es 95% lowe and uppe highes pos e io densi y in e als om he model. Colo ed poin s indica e
equency es ima es om he aw da a binned in o biweekly in e als. Selec ion mani es s as s aigh lines on
he logi scale o he y-axis whe e he slope co esponds o he s eng h o selec ion pe clade. Fo ecas s show
MLR model p edic ions a biweekly s eps up o 1 yea in he u u e. View on Nex s ain.
Figu e 21. G ow h ad an ages pe clade by egion as es ima ed by a MLR model. The “hie a chical”
ad an age ep esen s he global a e age i ness pe clade. The dashed e ical line a x=1 indica es he g ow h
ad an age o he pi o clade. Clades wi h g ow h ad an ages g ea e han 1 ha e a highe i ness han he pi o .
Colo ed poin s indica e he median g ow h ad an age pe clade and loca ion. E o ba s indica e he 95% lowe
and uppe highes pos e io densi y in e als om he model. View on Nex s ain.
23
An igenic p ope ies om e e and indi idual human se ology
[Raw e e da a edac ed]
Agg ega ing all cell-passaged HI da a wi h a i e model, we ind he highes an igenic ad ance is
associa ed wi h ecen C.3.1 s ains (Fig. 22). We obse e modes ad ance in C.5 subclades ca ying
HA1 T199A, P241K, and A202V. We obse e he highes ad ance in C.5 subclades ca ying HA1
L144P which has an es ima ed e ec o 1.4 log2 uni s, bu hese i uses emain a e. We also i a i e
model o indi idual human HI i e s om he VCM se ology wo king g oup. This model shows ha
an igenic ad ance o C.3.1 was high due o i s HA1 D197N subs i u ion (Fig. 23). Va ian s wi hin
C.5.1, C.5.6, and C.5.7 had simila an igenic ad ances, oo. Among hose subclades, we obse ed
ecu en an igenic e ec s o HA1 A154E (0.57
log2
uni s), D197N (0.51
log2
uni s), A202V (0.36
log2
uni s), and T199A (0.16 log2uni s).
Figu e 22. An igenic ad ance pe clade es ima ed om e e -based HI da a [5] o samples
collec ed since Feb ua y 1, 2025.
24
Figu e 23. An igenic ad ance pe clade es ima ed wi h a i e subs i u ion model [5] om human-
based HI da a om he VCM se ology wo king g oup.
25
