ARTICLE
In e e hnic analyses o blood p essu e loci in
popula ions o Eas Asian and Eu opean descen
Fumihiko Takeuchi1,2, Masa o Akiyama3, Nana Ma oba3, Tomohi o Ka suya4,5, Masahi o Naka ochi6,
Yasuha u Taba a e al.
#
Blood p essu e (BP) is a majo isk ac o o ca dio ascula disease and mo e han 200
gene ic loci associa ed wi h BP a e known. He e, we pe o m a mul i-s age genome-wide
associa ion s udy o BP (max N=289,038) p incipally in Eas Asians and me a-analysis in
Eas Asians and Eu opeans. We epo 19 new gene ic loci and ances y-specific BP a ian s,
con o ming o a common ances y-specific a ian associa ion model. A 10 unique loci,
dis inc non- a e ances y-specific a ian s colocalize wi hin he same linkage disequilib ium
block despi e he significan ly disco dan e ec s o he p oxy sha ed a ian s be ween he
e hnic g oups. The genome-wide anse hnic co ela ion o causal- a ian e ec -sizes is
0.898 and 0.851 o sys olic and dias olic BP, espec i ely. Some o he ances y-specific
associa ion signals a e also influenced by a selec i e sweep. Ou esul s p o ide new e i-
dence o he ole o common ances y-specific a ian s and na u al selec ion in e hnic
di e ences in complex ai s such as BP.
Fumihiko Takeuchi, Masa o Akiyama, Nana Ma oba, Tomohi o Ka suya, Masahi o Naka ochi,
Yasuha u Taba a e al.
#
DOI: 10.1038/s41467-018-07345-0 OPEN
Co espondence and eques s o ma e ials should be add essed o N.K. (email: [email p o ec ed]).
#
A ull lis o au ho s and hei a filia ions appea s a
he end o he pape .
NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.com/na u ecommunica ions 1
1234567890():,;
High blood p essu e is a majo isk ac o o ca dio ascula
diso de s such as co ona y hea disease and s oke.
App oxima ely 10 million dea hs each yea can be
a ibu ed o high blood p essu e globally1,2. An indi idual’s isk
o high blood p essu e is de e mined by gene ic, en i onmen al
and demog aphic ac o s and hei in e ac ion. Genome-wide
associa ion s udies (GWASs) and/o la ge-scale analyses by gene-
cen ic (o exome) a ia ion a ays ha e iden ified o e 200
gene ic loci influencing blood p essu e in p edominan ly
Eu opean-descen popula ions (hence o h e e ed o as Eu -
opeans)3–8. The p e alence o high blood p essu e is inc eased in
people o Eas Asian ances y, con ibu ing o hei inc eased isk
o s oke9. The easons o such e hnic di e ences emain o be
cla ified om he iewpoin o gene ic suscep ibili y as well as
li es yle. Al hough he ecen p og ession o GWAS in Eas
Asians allows us o make a p elimina y compa ison o associa ion
signals be ween he popula ions10,11, he sample sizes o GWAS
in Eas Asians ha e been gene ally much smalle han hose in
Eu opeans and unde -powe ed o he comp ehensi e in e e hnic
compa ison a a genome-wide scale. The e o e, la ge-scale gen-
ome-wide associa ion da a in bo h e hnic g oups a e equi ed o
sys ema ic, genome-wide in e e hnic compa ison.
He e, we pe o m a mul i-s age GWAS wi h a disco e y sample
o 130,777 Eas Asian indi iduals and ollow-up me a-analyses
in ol ing Eas Asians and Eu opeans (max N=289,038), o seek
bo h anse hnic and ances y-specific gene ic e ec s o fi e
blood p essu e pheno ypes: sys olic blood p essu e (SBP), dia-
s olic blood p essu e (DBP), pulse p essu e (PP), mean a e ial
p essu e (MAP), and hype ension. We hen seek in e e hnic
gene ic he e ogenei y o GWAS esul s be ween Eas Asians and
Eu opeans, ollowed by examina ion o na u al selec ion as a
po en ial mechanism unde lying he e hnic di e ences in gene ic
suscep ibili y o blood p essu e as well as o he complex ai s.
We epo ances y-specific blood p essu e a ian s and selec ion
signals in his s udy.
Resul s
Genome-wide associa ion analysis and lookup o eplica ion.
Adop ing a join analysis s a egy12, we pe o med a GWAS,
which consis ed o s age 1 (disco e y) and s age 2 ( ollow-up),
and a eplica ion s udy (Supplemen a y Fig. 1). In s age 1 o
GWAS, we used genome-wide associa ion da a om 130,777
indi iduals o Japanese ances y. Cha ac e is ics o pa icipan s,
geno yping a ays, and impu a ion a e summa ized in Supple-
men a y Tables 1, 2. Genomic con ol and in e cep s om linkage
disequilib ium (LD) sco e eg ession13 we e calcula ed a each
s udy le el (λ
GC
=0.89–1.24 and LD Sco e eg ession in e cep =
0.94–1.06), indica ing no esidual con ounding biases such as
popula ion s a ifica ion (Supplemen a y Table 2). Since he LD
Sco e eg ession in e cep can accoun o polygenic e ec s and
infla ion due o la ge sample size13, we applied he LD Sco e
eg ession in e cep as a co ec ion ac o o coho s wi h a
sample size o >3000 indi iduals (BBJ in his s udy). Genomic
con ol λ
GC
was used as a co ec ion ac o in he o he s udies.
Quan ile−quan ile plo s o each o he fi e blood p essu e ai s
a e p esen ed in Supplemen a y Fig. 2. Pheno ype-specific me a-
analysis was ca ied ou in he wo-s age app oach o bo h he
Eas Asian-specific and anse hnic me a-analyses (Supplemen-
a y Figs. 1, 3). Genome-wide associa ion esul s in he s age-1
iden ified 13,003 SNPs wi h a P alue < 1.6×10–5agains any
blood p essu e pheno ype in Eas Asians. This se o 13,003 SNPs
(sen inel SNPs lis ed in Supplemen a y Da a 1) was ollowed up
in 53,008 Eas Asian indi iduals (s age 2). Addi ionally, hese
13,003 SNPs we e examined in he anse hnic s age wi h
pheno ype-specific esul s o Eu opeans (max N=105,253) om
he In e na ional Conso ium on Blood P essu e (ICBP) GWAS
(N=69,909)3and he In e na ional Genomics o Blood P essu e
(iGEN-BP) Conso ium (N=35,344)10; he e was no o e lap in
samples be ween he wo da a se s. Sen inel SNPs (smalles P
alue agains any blood p essu e pheno ype) ha (i) eached P<
5×10–8in combined me a-analysis o s ages 1 and 2 and (ii)
showed e idence o suppo (P< 0.05) in he s age 2 me a-
analysis alone a e epo ed as no el loci in his s udy. We iden-
ified 19 p e iously un epo ed loci; 15 loci in Eas Asian-specific
analyses and 4 addi ional loci in he anse hnic me a-analysis
(Table 1and Supplemen a y Da a 2). By lookup in an indepen-
den eplica ion sample o Eu opeans om he UK Biobank (N=
422,771)14 plus Eas Asians om he China Kadoo ie Biobank (N
=94,201)15, we examined associa ions a ou lis o 19 sen inel
SNPs. Wi h he excep ion o ou SNPs, 15 sen inel SNPs showed
significan (P< 0.00263 =0.05/19) blood p essu e associa ion
wi h he conco dan di ec ion o allelic e ec s (Supplemen a y
Da a 2), hus alida ing he loci.
Regional associa ion plo s a e shown o he 19 newly iden ified
loci in Supplemen a y Fig. 4. Associa ions o he 19 sen inel SNPs
wi h o he blood p essu e pheno ypes a e demons a ed in
Supplemen a y Da a 3. In he disco e y s age, we also eplica ed
blood p essu e associa ions a p e iously epo ed loci, which
included 36 loci a genome-wide significance and u he 179 loci
a nominal significance (P< 0.05) (Supplemen a y Da a 4).
Func ional anno a ions o new loci. To iden i y candida e genes
a he newly iden ified blood p essu e loci, we examined whe he
any o he associa ion signals (sen inel blood p essu e SNP and
SNPs in Eas Asian LD 2> 0.80) we e coding o associa ed wi h
gene exp ession and o he ai s. A h ee loci, he sen inel SNPs
we e nonsynonymous, and 4 o 19 no el loci con ained SNPs (in
LD o 2> 0.80 wi h he op eVa ian ) associa ed wi h exp ession
quan i a i e ai loci (eQTLs) in a leas one issue in he
Geno ype-Tissue Exp ession (GTEx) da abase (Supplemen a y
Tables 3–5). A wo candida e gene loci, p oxy SNPs ( s760077 a
MTX1 and s3825942 a LOXL1) we e nonsynonymous and
associa ed wi h eQTLs. Fu he mo e, se en sen inel SNPs and/o
hei p oxy SNPs ( 2≥0.95) we e p e iously epo ed o be sig-
nifican ly associa ed wi h non-blood p essu e ai s (Supple-
men a y Da a 5), including a sen inel SNP ( s11642015) a he
FTO locus on 16q22, whose p oxies ( 2=0.97–0.99) ha e been
epo ed o associa e wi h body mass index and ype 2 diabe es16.
In ou s udy, s11642015 was significan ly associa ed wi h SBP,
MAP, and PP (P=1.9×10–12–1.3×10–9) wi h consis en ep o-
ducibili y in bo h s ages o Eas Asian analyses (Supplemen a y
Da a 1, 3). In addi ion, s11642015 was ecen ly iden ified o be
significan ly associa ed wi h SBP in mul i-ances y GWAS me a-
analysis inco po a ing gene−smoking in e ac ion17.
In e e hnic he e ogenei y o GWAS esul s. In he p esen
s udy, he a ailabili y o genome-wide associa ion da a om
>100,000 indi iduals o bo h Eas Asians and Eu opeans sepa-
a ely mo i a ed us o pe o m addi ional analyses o sys ema ic,
genome-wide in e e hnic compa ison. We used anse hnic
associa ion summa y s a is ics a ailable o bo h Eas Asian (N=
158,645 om s age 1 and iGEN-BP) and Eu opean (max N=
105,253 om ICBP and iGEN-BP) GWAS esul s in he sub-
sequen analysis o in e e hnic he e ogenei y. We defined in e -
e hnic he e ogenei y as he e ogenei y o gene ic (o allelic) impac
on SBP be ween he e hnic g oups. Using GWAS da a se s, we
compa ed he gene ic impac a anse hnic SNPs and de ec ed a
o al o eigh in e e hnic he e ogenei y loci— wo significan (P<
5×10–8) and six sugges i e (5×10–8≤P< 1×10–6) loci (Fig. 1a and
Supplemen a y Da a 6). In his s udy we dis inguished he allelic
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0
2NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions
impac om allelic e ec -sizes as p e iously defined by B own
e al.18; allelic impac is he geno ype−pheno ype co ela ion
coe ficien , which is app oxima ely a p oduc o allelic e ec and
mino allele equency (MAF). Se en o he eigh loci wi h
in e e hnic he e ogenei y we e anno a ed o he p e iously
epo ed blood p essu e loci; sen inel blood p essu e SNPs a hal
o hem (i.e., ou loci nea he CACNB2,C10o 107,SH2B3 and
DPEP1 genes3,5) we e ound o be in LD ( 2≥0.2) wi h he SNPs
showing some e idence o in e e hnic he e ogenei y. The wo
loci wi h significan in e e hnic he e ogenei y we e on 12q24 and
10q21 and bo h con ained mul iple associa ion signals (Fig. 1b).
Fo he egion on 12q24 spanning 1.5 Mb, wo independen
associa ion signals, each specific o Eu opeans (nea s3184504
a SH2B3) and Eas Asians (nea s671 a ALDH2), had been
iden ified19. We ound ha bo h o he signals we e esponsible
o he disco dan di ec ion o allelic e ec s on 12q24 (Fig. 1c and
Supplemen a y Fig. 5a, b). Simila ly, we obse ed wo indepen-
den associa ion signals nea he C10o 107 ansc ip on 10q21.2
(Fig. 1c and Supplemen a y Fig. 5c, d). The de i ed alleles
o ances y-specific sen inel SNPs on 10q21 ( s4590817 and
s145193831 specific o Eu opeans3and Eas Asians espec i ely)
a ose om a haplo ype sha ed be ween e hnic g oups, con aining
mul iple anse hnic SNPs. The disco dan di ec ion o e ec s
o he sha ed haplo ypes could be explained by al e na ion o
e ec s a ibu able o he de i ed alleles o s4590817 (dec easing
in Eu opeans) and s145193831 (inc easing in Eas Asians)
(Supplemen a y Fig. 6 and Supplemen a y Da a 7).
Ances y-specific SNP loci. A o al o 750 p e iously epo ed
SNPs (lis ed in Supplemen a y Da a 4) plus 19 newly iden ified
SNPs could be classified in o 485 loci by ega ding wo SNPs a
mos 500 kb apa o belong o he same locus. A e exclusion o
39 loci (MAF < 0.01 in bo h Eas Asians and Eu opeans, o no
da a a ailable in GWAS da a se s o bo h popula ions), 446 loci
we e e ained and ca ego ized in o wo g oups—g oup 1 and
g oup 2. G oup 1 consis ed o 382 loci wi h MAF ≥0.01 in bo h
popula ions and g oup 2 consis ed o 64 loci wi h po en ial e hnic
specifici y, i.e., MAF < 0.01 in ei he Eas Asians o Eu opeans.
G oup 2 was u he classified in o g oup 2a (46 loci wi h MAF <
0.01 in one popula ion and MAF ≥0.05 in he o he ) and g oup
2b (18 loci wi h MAF < 0.01 in one popula ion and 0.01 ≤MAF <
0.05 in he o he ) (Supplemen a y Fig. 7).
Wi h ega ds o in e e hnic he e ogenei y o associa ion
signals, we assumed wo dis inc scena ios: whe he he unde -
lying causal a ian s a e sha ed be ween he e hnic g oups o no .
Howe e , due o subs an ial in e e hnic di e ences in LD
s uc u e, i is no always easible o dis inguish be ween he
wo. Fi s , as an example o he po en ial nonsha ed causal a ian
(o ances y specifici y), we examined in e e hnic compa abili y
o gene ic impac on blood p essu e a 48 loci (46 loci in g oup 2a
plus 2 a ge loci wi h po en ial ances y specifici y—C10o 107
and CACNB2—included in g oup 1; Supplemen a y Fig. 7), whe e
sen inel common (MAF ≥0.05) blood p essu e SNPs o iginally
epo ed in a gi en e hnic g oup we e monomo phic o MAF <
0.01 in he second e hnic g oup3–8,19. Then, we in es iga ed
in e e hnic he e ogenei y a non- a e (MAF ≥0.01 in bo h e hnic
g oups) blood p essu e loci (g oup 1 in Supplemen a y Fig. 7)
ha migh be sha ed be ween he e hnic g oups as desc ibed la e .
Conside ing he obse a ions on 12q24 and 10q21, we explo ed
common p oxy SNPs o ming a haplo ype sha ed be ween e hnic
g oups a he locus (deno ed as haplo-SNPs), o which he mos
significan in e e hnic he e ogenei y o gene ic impac was
de ec ed (Supplemen a y Fig. 8a–c). A a o al o 11 loci (o 10
unique loci when he ALDH2 and SH2B3 loci on 12q24 we e
combined) (Supplemen a y Figs. 6, 9 and Supplemen a y Da a 7),
haplo-SNPs showed significan (P< 1.5×10–4unde egion-wise
co ec ion) he e ogenei y be ween wo e hnic g oups. A 8 o 11
loci, we ound ha dis inc common ances y-specific a ian s
colocalized wi hin he same LD block and ha he di ec ion o
e ec s o he p oxy sha ed SNPs was disco dan be ween he
e hnic g oups, simila o 12q24 and 10q21. On 5q14, o ins ance,
a genome-wide significan associa ion o s112862634 wi h SBP,
DBP, and MAP was de ec ed in Eas Asians o his s udy
(Supplemen a y Da a 1), while SBP associa ion o s10059921
was p e iously epo ed in i s icini y (456 kb apa om
Table 1 Gene ic loci newly iden ified o be associa ed wi h blood p essu e
Sen inel SNP Ch Posi ion EA/NEA EAF T ai NE ec P
Genome-wide significan and eplica ed
s2990220 1 155,190,254 A/T 0.83 MAP 183,654a−0.41 (0.06) 2.2×10−12
s6772151 3 46,896,499 A/C 0.29 DBP 156,503a0.28 (0.05) 7.8×10−9
s17622152 3 183,520,112 A/G 0.47 MAP 183,759a−0.25 (0.04) 2.0×10−8
s12209106 6 1,621,042 T/G 0.68 DBP 160,436a0.28 (0.05) 6.4×10−9
s78399431 7 1,141,470 A/G 0.24 MAP 179,411a0.30 (0.05) 9.6×10−9
s2125067 10 48,434,420 C/G 0.12 SBP 179,003a0.60 (0.10) 4.8×10−9
s2305013 11 120,340,060 A/T 0.85 SBP 180,894a−0.59 (0.09) 5.6×10−10
s5006548 12 32,692,233 T/G 0.16 HT 71,847a0.09 (0.02) 2.2×10−8
s1535464 14 100,793,431 A/G 0.10 SBP 183,690a−0.61 (0.10) 3.5×10−9
s66978877 19 18,455,657 T/C 0.55 HT 68,850a0.07 (0.01) 4.5×10−9
s6021247 20 50,108,980 A/G 0.58 SBP 183,785a0.37 (0.06) 5.0×10−9
s3853476 5 141,817,754 A/G 0.58 MAP 244,831b−0.20 (0.03) 6.0×10−9
s10821808 10 62,390,646 A/G 0.58 SBP 288,917b−0.29 (0.05) 3.4×10−9
s4418728 10 94,839,724 T/G 0.62 DBP 256,118b−0.20 (0.03) 1.5×10−8
s1078967 15 74,222,987 T/C 0.15 SBP 265,280b0.42 (0.07) 5.6×10−9
Genome-wide significan bu no eplica ed
s2076460 1 27,972,058 C/G 0.30 SBP 174,846a−0.42 (0.07) 3.6×10−9
s11642015 16 53,802,494 T/C 0.21 SBP 174,917a0.58 (0.08) 1.9×10−12
s9303509 17 64,530,887 A/C 0.40 SBP 183,769a0.37 (0.06) 3.9×10−9
s66658258 20 61,462,502 C/G 0.58 DBP 164,638a0.28 (0.05) 1.0×10−8
Posi ion is Build 37; EA: e ec allele; NEA: non-e ec allele; EAF: e ec allele equency; N: sample size (aEas Asians only; bwi h Eu opean ollow-up samples); E ec : as uni change in blood p essu e
(SE) pe e ec allele copy (SBP, DBP, PP, MAP) o as log odds a io pe e ec allele (HT)
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0 ARTICLE
NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions 3
a
–log10(P)
0.00
0.01
0.02
0.03
Gene ic
impac in EUR
0.00
0.01
0.02
0.03
Gene ic
impac in EUR
0.1
0.2
0.3
0.4
0.5
0.6
2 o
2 o
s4590817
10q21
EUR s4590817
0.1
0.2
0.3
s3184504
12q24
EUR s3184504
b
Gene ic impac in EAS
2 o
0.2
0.4
0.6
s145193831
s145193831
Posi ion on Ch 10 (Mb)
63.4 63.5 63.6
EAS
–0.02
0.00
0.02
Gene ic impac in EAS
Posi ion on Ch 12 (Mb)
2 o
s671
s671
112.0 112.5 113.0
EAS
–0.02
0.00
0.02
0.2
0.3
0.1
12q24
c
H3 H4 H1 H2 H9 H6 H7 H8
H3 H4 H1 H2 H9 H6 H8H5
Ances al haplo ype
Ances al haplo ype
s3184504_T
EAS
EUR
s671_A
(ALDH2)
H3 H2 H6 H7 H5 H4
H3 H2 H6 H8 H7 H4
Ances al haplo ype
Ances al haplo ype
s4590817_C
s145193831_T
EAS
EUR
10q21
(SH2B3)
20
15
10
5
0
123456789
Ch omosome
11 13 15 18 22
Fig. 1 In e e hnic he e ogenei y o gene ic impac o SBP. aManha an plo showing esul s o genome-wide scan o gene ic impac he e ogenei y. The
gene ic impac a anse hnic SNPs we e compa ed be ween wo popula ions o di e en ances ies using GWAS da a se s. bRegional plo s on 12q24 and
10q21, whe e he e we e mul iple SNPs wi h significan (P< 5×10–8) e idence o in e e hnic he e ogenei y (see Supplemen a y Da a 6). Bo de ed ci cles
ep esen SNPs wi h significan in e e hnic he e ogenei y. T anse hnic SNPs we e plo ed in wo panels a each locus; gene ic impac s o each SNP a e
deno ed sepa a ely o Eu opeans (EUR, op panel) and Eas Asians (EAS, bo om panel) on 12q24 (le ) and 10q21 ( igh ) such ha gene ic impac s in
Eu opeans a e posi i e. In he indi idual egional plo s, he co ela ion o ances y-specific sen inel SNP o o he SNPs a he locus is shown on a scale om
minimal (blue) o maximal ( ed); he sen inel SNPs hus benchma ked a e s3184504 (EUR specific) and s671 (EAS specific) on 12q24 and s4590817
(EUR specific) and s145193831 (EAS specific) on 10q21. The posi ion o ances y-specific sen inel SNP is indica ed by an a ow head. cPhylogene ic
ela ionships o ances y-specific sen inel SNPs wi h anse hnic haplo ypes de ec able in Eu opeans ( op) and Eas Asians (bo om) on 12q24 (le ) and
10q21 ( igh ). Each node co esponds o a haplo ype and he SNPs appea on he edges. The edge wid h eflec s he haplo ype equency in he
co esponding e hnic g oups. A each locus, blood p essu e inc easing and dec easing haplo ypes and de i ed, ances y-specific alleles a e colo ed in ed
and blue, espec i ely
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0
4NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions
s112862634) in Eu opeans8. I u ned ou ha s112862634 was
in s ong LD (Eas Asian LD 2=0.95) wi h a haplo-SNP
( s6882046) a his locus and dis inc common ances y-specific
a ian s wi h mu ually in e ed gene ic e ec s—Eu opean-
specific s10059921 (MAF =0.09 in EUR) and Eas Asian-
specific s78245349 (MAF =0.46 in EAS)—did colocalize in his
egion (Supplemen a y Fig. 8a and Supplemen a y Da a 7). A he
emaining 3 (o 11) loci, al e na e a e ances y-specific SNPs
we e likely o exis in he second e hnic g oup, al hough hey we e
no de ec able in ou sea ch o public da abases. We designa ed
hese as a common ances y-specific a ian associa ion model as
discussed below.
We hypo hesized ha he e we e h ee majo combina ions o
Eas Asian-/Eu opean-specific SNPs and hei esul an di ec ion
o e ec s o haplo-SNPs o ming a sha ed haplo ype a he locus,
as schema ically shown in Supplemen a y Fig. 6b. In acco dance
wi h his no ion, we de ec ed h ee ypes in his s udy
(Supplemen a y Fig. 9), among which he fi s and majo ype
(32 o 48 loci in g oup 2a) consis ed o he cases wi h mu ually
in e ed gene ic e ec s as explained abo e. The second ype
consis ed o hose wi h dis inc ances y-specific a ian s showing
conco dan di ec ions o e ec such as he FGR locus (Supple-
men a y Figs. 8b, 9). The hi d ype consis ed o hose wi h
dis inc ances y-specific a ian s showing disco dan gene ic
e ec s, one o which appea ed o be almos neu al such as he
GNAS/EDN3 locus (Supplemen a y Fig. 8c). Howe e , wi hou
using la ge sample sizes, i appea ed o be di ficul o show
s a is ically significan in e e hnic he e ogenei y in pa icula , o
hose in he second o hi d ype. A one locus (nea HSD17B1 on
17q21), a haplo-SNP could no be selec ed (Supplemen a y
Fig. 8d and Supplemen a y Da a 7), p esumably because o he
ances y-specific LD s uc u e and he modes s eng h o
associa ion in he index e hnic g oup (Eu opeans a he locus)
o his s udy.
Fo loci wi h po en ial ances y specifici y (i.e., MAF < 0.01 in
one popula ion and 0.01 ≤MAF < 0.05 in he o he ; 18 loci
classified as g oup 2b in Supplemen a y Fig. 7), we did no
in es iga e in e e hnic he e ogenei y o associa ion signals
because o di ficul ies in he ele an es o a e (MAF < 0.01)
and low- equency (0.01 ≤MAF < 0.05) gene ic a ian s by using
impu ed GWAS esul s20.
He e ogenei y a a ian s polymo phic in bo h ances ies.In
addi ion o he ances y-specific loci, we in es iga ed in e e hnic
he e ogenei y a non- a e (MAF ≥0.01 in bo h e hnic g oups)
blood p essu e loci ha migh be sha ed be ween he e hnic
g oups; 382 es ed loci we e ei he p e iously epo ed o newly
iden ified in he p esen s udy (deno ed as g oup 1 in Supple-
men a y Fig. 7a). Since ICBP and iGEN-BP (Eu opean) da a we e
impu ed wi h HapMap SNPs, app oxima ely one- hi d o g oup-1
SNPs we e una ailable in ou Eu opean GWAS da a se s. Thus,
242 (ou o 382) loci in g oup 1 we e subjec ed o in e e hnic
compa ison o gene ic impac on a lead blood p essu e ai
(Supplemen a y Da a 8). Al hough majo i y o hem appea ed o
show conco dan e ec s (co ela ion coe ficien =0.754), nine
sen inel SNPs (3.7%) showed significan (P
he e o
< 2.1×10–4)
in e e hnic he e ogenei y (Supplemen a y Fig. 10). Gene ic
impac s we e mo e p ominen in Eu opeans han in Eas Asians
a eigh o nine loci apa om s1451538 in SLC28A1, a which
gene ic impac s we e p ominen in Eas Asians bu no in Eu -
opeans (Supplemen a y Da a 8). The e we e no p oxy SNPs nea
each o he eigh loci in he same LD block (Supplemen a y
Fig. 11), which could ha e shown s onge associa ion signals in
Eas Asians han he sen inel SNPs o iginally epo ed in Eu -
opeans due o po en ial in e e hnic di e ences in LD s uc u e, i
any. O no e is he finding on 10q23 nea PLCE1, he e was
ano he SBP associa ion signal a s7080472 in Eas Asians (P=
3.9×10–8in he combined samples; Supplemen a y Da a 1)
despi e he absence o p ominen associa ion a s932764, whose
associa ion was p e iously epo ed4and p ominen in Eu opeans
(Supplemen a y Fig. 11). s7080472 was loca ed in he LD block
nex o he one o s932764 (Eas Asian LD 2=0.003 be ween
s7080472 and s932764). On 10q21 nea C10o 107, a DBP
associa ion signal was p e iously epo ed a s153044021, which
we ound o be in LD (Eu opean LD 2=0.48) wi h an ances y-
specific SNP a he locus, s4590817, a o emen ioned (Supple-
men a y Da a 7). Also, on 10p12 nea CACNB2, a DBP associa-
ion signal was p e iously epo ed a s18133533, which we
ound o be in LD (Eu opean LD 2=0.56) wi h an ances y-
specific SNP a he locus, s12258967. These indica ed ha
in e e hnic he e ogenei ies iden ified o non- a e anse hnic
a ian s on 10q21 and 10p12 we e he cases o which common
ances y-specific a ian s we e ac ually esponsible.
By calib a ing he p opo ion in he g oup-1 subse , in which
blood p essu e GWAS esul s o in e e hnic compa ison we e
a ailable o 242 (o 382) loci, we es ima ed he p opo ion o loci
showing significan in e e hnic he e ogenei y wi hin he o al
blood p essu e loci es ed (N=446). The es ima ed p opo ion
was 2.5% each in g oup 1 and g oup 2a, whe e he C10o 107 and
CACNB2 loci we e coun ed in g oup 2a (Supplemen a y Fig. 7b).
Gene ic co ela ion and powe o GWAS. As an app oach o
quan i a i ely e alua ing he in e e hnic di e ences in blood
p essu e GWAS esul s, we es ima ed he gene ic co ela ion
using summa y s a is ics o he en i e spec um o GWAS asso-
cia ions18.Wefi s es ima ed he SNP-based he i abili y (h2)o
SBP and DBP (Fig. 2). Fo SBP, h2es ima es in ou s udy we e
0.107 (SE 0.007) o Eas Asians and 0.086 (SE 0.009) o Eu -
opeans and lowe han a p e iously epo ed UK Biobank es i-
ma e o 0.156 (SE 0.004)22 calcula ed by he momen -ma ching
me hod in Eu opeans. This disc epancy was likely due o he
me hodological di e ences in SNP-based he i abili y analyses
be ween he s udies bu does no appea o a ec gene ic-
co ela ion es ima es hemsel es23. Also, he h2o DBP was
almos compa able be ween he e hnic g oups in his s udy. Then,
we ound ha he gene ic co ela ions in SBP and DBP we e 0.898
(SE 0.040) and 0.851 (SE 0.046) espec i ely, and significan ly
di e en om 1 (P=0.005 o SBP and P=0.0007 o DBP).
This indica ed ha he allele-subs i u ion e ec -sizes di e ed
significan ly be ween he wo e hnic g oups despi e he epo edly
subs an ial gene ic o e lap in blood p essu e ai s (Supplemen-
a y Da a 4).
To es ima e he deg ee o in e e hnic o e lap and nono e lap
o blood p essu e loci, we u he calcula ed he powe o GWAS
o di e en sample sizes (i.e., 100K, 200K, and 500K) based on
he i abili y pa ame e s (see de ails in Supplemen a y Me hods)
ia modeling, compu ing and andom sampling (Fig. 3and
Supplemen a y Figs. 12, 13). Simila o Eu opeans, he ecen
p og esses o GWAS in Eas Asians p omp ed us o in es iga e
di e en sample sizes in p epa a ion o much-la ge anse hnic
me a-analysis. When GWASs o he same size we e ca ied ou
o SBP and DBP, i was expec ed ha an almos equi alen
numbe o genome-wide significan loci could be iden ified in
bo h Eas Asians and Eu opeans bu he numbe o o e lap was
less han hal .
We ex ended he in e e hnic analyses o o he complex ai s
such as plasma lipid le el, an h opome ic measu emen , and
ype 2 diabe es using published GWAS summa y s a is ics o
ela i ely la ge numbe o samples (Supplemen a y Table 6).
Al hough gene ic co ela ion appea ed o be a ied among he
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0 ARTICLE
NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions 5

complex ai s examined (Fig. 2), we ound ha he p opo ion o
nono e lap [(nono e lap) / (o e lap +nono e lap)] was ela-
i ely consis en ac oss he ai s o he same sample size;
0.71–0.82 o 100K, 0.65–0.78 o 200K and 0.46–0.70 o 500K
(Fig. 3and Supplemen a y Fig. 12). As he sample sizes in bo h
e hnic g oups become la ge , we can expec a highe p opo ion
o in e e hnic o e lap; ne e heless, mo e han o nea ly hal o
he genome-wide significan loci may no o e lap be ween he
e hnic g oups o GWAS o he same sample size.
Selec i e sweeps a ances y-specific loci. Subsequen ly, we c e-
a ed a lis o ances y-specific loci o SBP, DBP and o he
complex ai s in which an SNP- ai associa ion was genome-
wide significan in one e hnic g oup (e.g., Eas Asians) bu no
significan associa ion signal was de ec able in ano he (e.g.,
Eu opeans) due o low allele equency (MAF < 0.05) (Supple-
men a y Da a 9). Fo he loci wi h he same SNPs being mono-
mo phic in he second e hnic g oup, ou selec ion c i e ia o
ances y-specific loci could be ega ded s ingen in ha he
absence o locus-wide significan associa ion signals in he icini y
(≤500 kb) o he es ed SNPs was equi ed.
A la ge numbe o significan loci had been epo ed in
Eu opeans compa ed o Eas Asians, eflec ing he di e ences in
sample size o GWAS conduc ed o da e (mean o fi e ai s was
81,991 in Eas Asians s. 202,390 in Eu opeans) (Fig. 4and
Supplemen a y Fig. 14). Thus, he o al numbe o ances y-
specific loci ac oss he examined ai s was smalle in Eas Asians
(10 loci) han in Eu opeans (63 loci). While i was mos
p ominen o heigh , he sen inel SNPs a he ances y-specific
loci ended o ha e bo h lowe MAF (0.20 ± 0.04 in Eas Asians,
0.16 ± 0.01 in Eu opeans) and gene ic impac (0.020 ± 0.002 in
Eas Asians, 0.014 ± 0.0004 in Eu opeans) ac oss he ai s.
Among a lis o ances y-specific loci o mul iple ai s, we
iden ified e idence o a posi i e selec ion a fi e unique loci using
a highly sensi i e algo i hm, haploPS24 (Fig. 5and Supplemen a y
Da a 9). Fo blood p essu e, a sen inel SNP s56174355 on 17q23
p e iously epo ed o be associa ed wi h DBP only in Eu opeans8
was localized o a egion wi h e idence o posi i e selec ion in
Eas Asians (Fig. 5d). In his egion, we obse ed he long
haplo ypes a high equencies (i.e., 70–80%) o be selec ed
exclusi ely in Eas Asians, on which he p esen -day majo allele
(G o s56174355) could eside, whe eas he mino allele (T o
s56174355) was associa ed wi h lowe DBP in Eu opeans. Thus,
a selec i e sweep in he egion is conside ed o ha e e ained he
majo allele ha was likely beneficial in he popula ions o Eas
Asian ances y; con e sely, his has educed MAF in Eas Asians
(T allele: 0.03 in Eas Asians s. 0.10 in Eu opeans). We ound
simila examples o he ai s o he han blood p essu e in ou
egions: s12748152 o LDL-C and iglyce ides, s17031005 o
T2D, s11862222 o heigh and s4253772 o o al choles e ol
(Fig. 5a−c, e). The e was a significan (P
Binomial
=4.2 × 10–5)
inc ease in he incidence o ecen selec ion signals a he
ances y-specific loci, gi en ha a o al o 405 dis inc genomic
egions we e iden ified o show e idence o posi i e selec ion
ac oss 14 popula ions wo ldwide24.
Discussion
Ou GWAS in 183,785 Eas Asian indi iduals iden ified 15 new
gene ic loci influencing blood p essu e pheno ypes and 4 addi-
ional loci when combined wi h Eu opean indi iduals (max N=
289,038) (Table 1). O he 19 newly iden ified loci, 15 loci we e
eplica ed in an independen sample o Eu opeans (N=422,771)
plus Eas Asians (N=94,201) (Supplemen a y Da a 2). A no able
ea u e o his s udy is he use o a ela i ely la ge disco e y-s age
sample size in popula ions o non-Eu opean descen , he eby
enabling us o iden i y a numbe o gene ic loci ha ha e no been
epo ed by GWAS me a-analysis in Eu opeans (Fig. 3). By
combining he Eas Asian da a wi h Eu opean da a, we we e also
able o seek in e e hnic gene ic he e ogenei y o GWAS esul s o
blood p essu e be ween he wo ances ies (Fig. 1) as well as o he
complex ai s. In pa icula , he p esen s udy p o ides examples
o in e e hnic gene ic he e ogenei y, al hough he incidence may
no be high, disco e ing wo ema kable phenomena: (1) he
colocaliza ion o dis inc ances y-specific a ian s ha a e no
a e and can exe mu ually in e ed gene ic e ec s be ween he
e hnic g oups and (2) he po en ial in ol emen o na u al
selec ion in he occu ence o ances y-specific associa ion signals.
Among gene ic loci iden ified in Eas Asians, o no e is he
finding ha a wo loci on 1p35 and 3p21, he la e o which
esides nea he associa ion signal p e iously epo ed in Chi-
nese25, sen inel SNPs ( s2076460 and s3774447) appea o be
specific o Eas Asians; i.e., in Eu opeans he co esponding SNPs
we e monomo phic and no significan associa ion signals we e
de ec able in he icini y (Supplemen a y Da a 7). These suppo
he possible p esence o mul iple Eas Asian-specific associa ions
as well as Eu opean-specific ones.
0.00
0.25
0.50
0.75
1.00
T ans-ances y gene ic co ela ion (95% CI)
EAS s. EUR
SBP
DBP
HDL
LDL
TC
TG
T2D
BMI
Heigh
He i abili y (95% CI)
0.00
0.25
0.50
0.75
1.00
EAS
SBP
DBP
HDL
LDL
TC
TG
T2D
BMI
Heigh
He i abili y (95% CI)
0.00
0.25
0.50
0.75
1.00
EUR
SBP
DBP
HDL
LDL
TC
TG
T2D
BMI
Heigh
Fig. 2 T anse hnic gene ic co ela ion and SNP-based he i abili y. SNP-based he i abili y o SBP, DBP and o he complex disease and pheno ype ai s is
shown sepa a ely o Eas Asians (EAS) and Eu opeans (EUR) by using he published GWAS summa y s a is ics (Supplemen a y Table 6). The whiske s
a e 95% confidence in e als o each alue
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0
6NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions
I has been sugges ed ha GWAS signals a e p oduced by
causal a ian s ha a e common and sha ed be ween ances y
g oups26, wi h e idence o al e na i e a e a ian associa ion
models (e.g., syn he ic associa ion27) ha a e assumed o be
es ic ed o a limi ed numbe o loci. Apa om hese models,
we ha e disco e ed a new model in which gene ic e ec s o
anse hnic SNPs ha o m a sha ed haplo ype a a locus a e
d i en by causal a ian s ha a e ances y-specific bu a e no
a e, which can be called a common ances y-specific a ian
associa ion model. We p e iously epo ed on 12q24 he Eas
Asian-specific associa ion signal a ALDH2 wi h blood p essu e,
which was loca ed nea he associa ion signal a SH2B3 iden ified
–0.02 0.00 0.02
–0.02
0.00
0.02
= 0.12
Signi ican in NEUR = 105,253
Signi ican in NEAS= 158,645
DBP
a
S anda dized e ec -size
in EUR
S anda dized e ec -size in EAS
–0.2 –0.1 0.0 0.1 0.2
–0.15
–0.10
–0.05
0.00
0.05
0.10
0.15
= 0.05
Signi ican in NEUR = 173,082
Signi ican in NEAS= 31,732
LDL-C
S anda dized e ec -size
in EUR
S anda dized e ec -size in EAS
–0.04 0.00 0.04
–0.04
–0.02
0.00
0.02
0.04
Signi ican in NEUR = 158,186
Signi ican in NEAS= 25,066
= 0.07
T2D
S anda dized e ec -size
in EUR
S anda dized e ec -size in EAS
–0.050 –0.025 0.000 0.025 0.050
–0.050
–0.025
0.000
0.025
0.050
Signi ican in NEUR = 322,154
Signi ican in NEAS= 158,284
= 0.26
BMI
S anda dized e ec -size
in EUR
S anda dized e ec -size in EAS
Numbe o SNPs in bin
110
3106
–0.050 –0.025 0.000 0.025 0.050
–0.06
–0.03
0.00
0.03
0.06
Signi ican in NEUR = 253,279
Signi ican in NEAS= 36,227
= 0.20
Heigh
S anda dized e ec -size
in EUR
S anda dized e ec -size in EAS
100K 200K 500K
21 59 224
22 8 13 18
62 13 25 45
234 18 43 110
100K200K
500K
Numbe o loci
de ec able in a single
EUR GWAS, NEUR=
Numbe o loci
de ec able in a single
EAS GWAS, NEAS =
b
28 66 198
59 17 28 43
137 21 41 78
383 26 55 135
100K200K
500K
Numbe o loci
de ec able in a single
EUR GWAS, NEUR=
100K 200K 500K
Numbe o loci
de ec able in a single
EAS GWAS, NEAS =
17 56 233
9579
32 9 17 27
144143787
100K 200K 500K
Numbe o loci
de ec able in a single
EAS GWAS, NEAS =
100K200K
500K
Numbe o loci
de ec able in a single
EUR GWAS, NEUR=
26 94 403
13 6 9 12
51 13 26 43
249 21 61 155
100K 200K 500K
100K 200K 500K
Numbe o loci
de ec able in a single
EAS GWAS, NEAS =
100K200K
500K
Numbe o loci
de ec able in a single
EUR GWAS, NEUR=
93 239 656
153 56 95 136
370 76 157 286
836 90 220 525
Numbe o loci
de ec able in a single
EAS GWAS, NEAS =
100K200K
500K
Numbe o loci
de ec able in a single
EUR GWAS, NEUR=
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0 ARTICLE
NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions 7
in Eu opeans19. We also epo ed ha hese wo associa ion
signals we e phylogene ically independen , al hough a dis ance
be ween he sen inel SNPs ( s671 and s3184504) was ela i ely
close (357 kb apa ). Mo eo e , in he p esen s udy we ha e
de ec ed a numbe o anse hnic SNPs in he 12q24 egion o
show highly significan he e ogenei y o gene ic impac on SBP
be ween he e hnic g oups (e.g., β
EAS
=−0.73 and β
EUR
=0.37,
P
he
=9.74×10–21 a s4766566), whe e in e ed gene ic e ec s
a e a ibu able o each o he ances y-specific sen inel SNPs; a
simila si ua ion was also obse ed on 10q21 o ep oduce his
phenomenon (Fig. 1and Supplemen a y Da a 6). Using ances y-
specific SNPs ha a e epo ed o each genome-wide significance
in ei he o he e hnic g oups, we ha e ound u he e idence
suppo ing he common ances y-specific a ian associa ion
model a 11 o 48 loci (23%) examined (Supplemen a y Da a 7).
This co esponds o 2.5% o o al non- a e genome-wide sig-
nifican blood p essu e loci epo ed o da e in popula ions o
Eu opean and/o Eas Asian descen 3–8,19 (Supplemen a y Fig. 7).
Al hough i is beyond he scope o his s udy, pa o he low-
equency a ian s a g oup-2b loci (which cons i u e 4.0% o he
es ed blood p essu e loci) may also be ances y-specific20. These
findings a e impo an and should be kep in mind in he wo
well-known applica ions o anse hnic GWAS, i.e., me a-analyses
o inc ease he powe o de ec ing new suscep ibili y loci and fine
mapping.
Wi h he inc ease in sample size used o GWAS me a-analysis,
i is expec ed ha a la ge numbe o gene ic loci will be de ec ed,
and he dis ibu ion o such loci in he genome will become
dense . Associa ion signals anno a ed o he same locus a e
empi ically defined such ha a se o SNPs a e bounded by
pai wise co ela ion wi h he index SNP o 2≥0.1−0.3 wi hin
±250–500 kb o he index SNP26,28. This is usually discussed in
he con ex o locus he e ogenei y a he han allelic he e o-
genei y. Apa om ex eme cases in which an index SNP is
monomo phic in he second e hnic g oup as abo e-men ioned,
he cases in which a common a ian in ques ion is less common
o e en a e in he second e hnic g oup necessi a e a g ea e
sample size o achie e compa able s a is ical powe o de ec ing a
significan associa ion. We should be ca e ul in se ing app o-
p ia e significance h esholds o main ain a balance be ween
gene a ing spu ious associa ions and missing ue modes asso-
cia ions in he second e hnic g oup. Hence, we chose ances y-
specific loci based on he P alue h esholds adjus ed o he
numbe o SNPs loca ed ≤500 kb om he sen inel SNP. This se
o loci may no exclude some cases wi h insu ficien s a is ical
powe bu can include he cases in which gene ic impac a he
locus is la gely ega ded as specific o he o iginal e hnic g oup.
E idence o posi i e selec ion was obse ed a fi e unique loci
among he lis o ances y-specific loci (Fig. 5and Supplemen a y
Da a 9).
In addi ion o ances y-specific loci, al hough he p opo ion
appea s o be ela i ely modes (app oxima ely 2.5%), we ha e
ound significan in e e hnic he e ogenei y o gene ic impac a a
numbe o blood p essu e loci ha a e non- a e in bo h
ances ies, wi h mos o hem o iginally epo ed in Eu opeans o
da e. I is assumed ha he po en ial p esence o modifie genes
and/o gene−en i onmen in e ac ions can con ibu e o such
in e e hnic he e ogenei y bu he o e all influences and unde -
lying mechanisms emain o be in es iga ed. When combined
wi h ances y-specific a ian associa ions (a g oup-2a o g oup-
2b loci in Supplemen a y Fig. 7), >5% o blood p essu e loci a e
likely o show significan in e e hnic he e ogenei y be ween Eas
Asians and Eu opeans.
Acco ding o ou SNP-based he i abili y analysis, he genome-
wide co ela ion o causal- a ian e ec -sizes a SNPs common in
bo h ances y g oups is 0.898 and 0.851 o SBP and DBP,
espec i ely (Fig. 2). Pa o he educed in e e hnic co ela ion is
a ibu able o anse hnic a ian s ha a e common ac oss
popula ions bu show subs an ial in e e hnic he e ogenei y,
al hough he p opo ion o such a ian s may no be high (e.g., 9
loci wi h in e e hnic he e ogenei y de ec ed in g oup-1; Supple-
men a y Fig. 7). E en hough hey a e no included in he SNP-
based he i abili y analysis, ances y-specific a ian s (a g oup-2a
loci in Supplemen a y Fig. 7) can influence he pe -allele e ec -
sizes o a numbe o anse hnic SNPs a he co esponding loci
ia LD, e.g., a he C10o 107 and CACNB2 loci.
In summa y, we iden i y a o al o 19 gene ic loci ha ha e no
been epo ed p e iously by GWAS me a-analysis, using ela-
i ely la ge disco e y-s age sample size in Eas Asian popula ions.
By compa ing GWAS da a o wo e hnic g oups, we ha e newly
defined, so o speak, a common ances y-specific a ian asso-
cia ion model, which should be b ough o a en ion in he
applica ions o anse hnic GWAS.
Me hods
Popula ions and geno yping. Desc ip ion o he s udy design and pheno ype
measu emen o each Eas Asian s udy (o coho ) pa icipa ing in GWAS me a-
analysis is p o ided in he Supplemen a y Me hods. Desc ip i e s a is ics o he
indi iduals, geno yping a ays, quali y con ol fil e s, and geno ype impu a ion
applied o he indi idual s udies a e p o ided in Supplemen a y Tables 1, 2; 1000
Genomes Phase 3 e e ence panel was used o impu a ion in all s udies excep BBJ
(1000 Genomes Phase 1) and TMM CommCoho S udy (ToMMo 2KJPN panel
plus 1000 Genomes Phase 3). SNP alleles we e o ien ed o he o wa d s and o he
GRCh37/hg19 e e ence sequence o he human genome. Collec ion o da a and
samples by he coho s pa icipa ing in he s udy was app o ed by espec i e
esea ch e hics commi ees, and w i en consen o pa icipa ion was p o ided by
all esea ch pa icipan s.
Pheno ype modeling and SNP associa ion analysis. Fo indi iduals aking
an ihype ensi e he apies, blood p essu e was impu ed by adding 15 mmHg and
10 mmHg o SBP and DBP alues, espec i ely. MAP and PP we e calcula ed as
MAP =(2 DBP +SBP)/3 and PP =SBP –DBP. In each s udy, he associa ion o
blood p essu e (SBP, DBP, MAP o PP) wi h SNP allele dose was es ed using
linea eg ession adjus ed o age, sex, and any s udy-specific co a ia es. Hype -
ensi e cases we e defined as ollows: (i) SBP ≥160 mmHg and/o DBP ≥100
mmHg and/o on an ihype ensi e ea men and (ii) age o onse ≤65 yea s.
No mo ensi e con ols we e defined as ollows: (i) SBP < 130 mmHg and DBP < 85
mmHg and no on an ihype ensi e ea men and (ii) age ≥50 yea s. In each s udy,
he associa ion o a dicho omous ai o hype ension s a us wi h SNP allele dose
was es ed using logis ic eg ession adjus ed o sex and any s udy-specific co -
a ia es. The e ec -sizes and s anda d e o s es ima ed in linea and logis ic
eg essions we e used in subsequen me a-analysis.
Fig. 3 Dis ibu ion o SNP e ec -size in GWAS and powe o GWAS. They a e compa ed be ween Eas Asians and Eu opeans o DBP, low-densi y
lipop o ein choles e ol (LDL-C), ype 2 diabe es (T2D), body mass index (BMI) and heigh . aDis ibu ion o SNP e ec -size in ac ual GWAS conduc ed in
Eas Asians (x-axis) and Eu opeans (y-axis). The e ec -size o an SNP was s anda dized such ha each o he ai and allele has a uni a iance. The
s anda dized e ec -size equals he gene ic impac . A posi i e e ec -size indica es a highe ai alue o he ALT allele compa ed o he REF allele o he
1000 Genomes (1000G) phase-3 da a se . The ho izon al and e ical ba s o he bo om and igh o he plo s indica e he ange o e ec -sizes, in which
genome-wide significan SNPs a e localized. bThe expec ed numbe s o genome-wide significan loci de ec able in a single GWAS and hei in e e hnic
o e lap. The numbe o SNPs was scaled o 1000G SNPs e en o GWAS in which HapMap-de i ed SNPs we e assayed. SNPs loca ed ≤500 kb we e
ega ded o be a he same suscep ibili y locus. The numbe s o loci we e in e ed om he he i abili y model shown in Supplemen a y Fig. 13, whe e ue
obse able e ec -sizes we e compu ed based on 100 ials o andom sampling unde he assumed he i abili y pa ame e s (see Me hods)
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0
8NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions
DBP
Sen inel SNPs o loci associa ed in EAS
Speci ic o EAS (1)
No speci ic (39)
0.00
0.01
0.02
0.03
0.04
S anda dized
e ec -size in EAS (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EAS
LDL-C Speci ic o EAS (0)
No speci ic (11)
0.00
0.05
0.10
0.15
0.20
S anda dized
e ec -size in EAS (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EAS
T2D
Speci ic o EAS (2)
No speci ic (22)
0.000
0.025
0.050
0.075
S anda dized
e ec -size in EAS (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EAS
BMI
Speci ic o EAS (6)
No speci ic (61)
0.00
0.01
0.02
0.03
0.04
0.05
S anda dized
e ec -size in EAS (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EAS
Heigh
Speci ic o EAS (0)
No speci ic (70)
0.00
0.02
0.04
0.06
S anda dized
e ec -size in EAS (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EAS
Sen inel SNPs o loci associa ed in EUR
Speci ic o EUR (8)
No speci ic (47)
0.00
0.01
0.02
0.03
0.04
S anda dized
e ec -size in EUR (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EUR
Speci ic o EUR (7)
No speci ic (65)
0.00
0.05
0.10
0.15
0.20
S anda dized
e ec -size in EUR (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EUR
0.000
0.025
0.050
0.075
Speci ic o EUR (3)
No speci ic (36)
S anda dized
e ec -size in EUR (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EUR
Speci ic o EUR (6)
No speci ic (68)
0.00
0.01
0.02
0.03
0.04
0.05
S anda dized
e ec -size in EUR (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EUR
0.00
0.02
0.04
0.06
Speci ic o EUR (21)
No speci ic (459)
S anda dized
e ec -size in EUR (absolu e alue)
0.0 0.1 0.2 0.3 0.4 0.5
Mino allele equency in EUR
Fig. 4 In e e hnic compa ibili y o GWAS esul s o DBP, LDL-C, T2D, BMI, and heigh . Each poin in he plo s ep esen s a sen inel SNP wi h genome-wide
significance in he GWAS summa y s a is ics (Supplemen a y Table 6), plo ed wi h i s s anda dized e ec -size (in y-axis) agains mino allele equency
(in x-axis) o Eas Asians (EAS in he le column) and Eu opeans (EUR in he igh column). SNPs specific o ei he o he e hnic g oups a e colo ed in ed;
ances y-specific associa ion was defined such ha he sen inel SNPs a he co esponding loci eached genome-wide significance (P< 5×10–8) in one
e hnic g oup bu we e non-polymo phic o a e (MAF < 0.05) in ano he e hnic g oup
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0 ARTICLE
NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions 9
Uni e si y Medical Cen e G oningen, Uni e si y o G oningen, Hanzeplein 1, 9713 GZ G oningen, Ne he lands.
113
Depa men o Obs e ics and
Gynecology, Uni e si y Hospi al o Oulu, Uni e si y o Oulu, Oulu FI-90014, Finland.
114
Depa men o Clinical Sciences/Obs e ics and Gynecology,
Uni e si y o Oulu, Oulu FI-90014, Finland.
115
Cen e o Ca dio ascula Diagnos ics and P e en ion, Cle eland Clinic, Cle eland, OH 44195, USA.
116
Depa men o Cellula and Molecula Medicine, Le ne Resea ch Ins i u e, Cle eland Clinic, Cle eland, OH 44195, USA.
117
Keck School o
Medicine, Uni e si y o Sou he n Cali o nia, Los Angeles, CA 90089, USA.
118
Resea ch Cen e o P e en ion and Heal h, Glos up Uni e si y
Hospi al, 2600 Glos up, Denma k.
119
Depa men o Epidemiology and Public Heal h, UCL, London WC1E 6BT, UK.
120
Hea Cen e, Depa men o
Ca diology, Tampe e Uni e si y Hospi al, and Uni e si y o Tampe e School o Medicine, FI-33521 Tampe e, Finland.
121
Depa men o Public Heal h
& Clinical Medicine, Sec ion o Family Medicine, Umeå uni e si e , SE-901 85 Umeå, Sweden.
122
Depa men o Epidemiology, Uni e si y Medical
Cen e G oningen, Uni e si y o G oningen, Hanzeplein 1, 9713 GZ G oningen, Ne he lands.
123
Depa men o Vascula Medicine, Uni e si y
Medical Cen e U ech , 3508 GA U ech , Ne he lands.
124
Alzheime Sco land Demen ia Resea ch Cen e, Uni e si y o Edinbu gh, 7 Geo ge
Squa e, Edinbu gh EH8 9JZ, UK.
125
Academic Sec ion o Ge ia ic Medicine, Ins i u e o Ca dio ascula and Medical Sciences, Facul y o Medicine,
Uni e si y o Glasgow, Glasgow G4 0SF, UK.
126
Ins i u e o Clinical Molecula Biology, Ch is ian-Alb ech s-Uni e si y o Kiel, Kiel 24105, Ge many.
127
Depa men o Ca diology, Leiden Uni e si y Medical Cen e , 2300 RC Leiden, Ne he lands.
128
Depa men o Clinical Physiology, Tampe e
Uni e si y Hospi al, FI-33521 Tampe e, Finland.
129
Depa men o Clinical Physiology, Uni e si y o Tampe e School o Medicine, FI-33014 Tampe e,
Finland.
130
Depa men o Obs e ics and Gynecology, Oulu Uni e si y Hospi al, PO Box 23FI-90029 Oulu, Finland.
131
Depa men o Medicine,
Tu ku Uni e si y Hospi al, FI-20521 Tu ku, Finland.
132
Depa men o Medicine, Uni e si y o Tu ku, FI-20014 Tu ku, Finland.
133
Du e Cen e o
Ca diogene ic Resea ch, ICIN-Ne he lands Hea Ins i u e, 3511 GC U ech , Ne he lands.
134
Ins i u e o Ca dio ascula Science, Facul y o
Popula ion Heal h Sciences, Uni e si y College London, London WC1E 6BT, UK.
135
Depa men o Nu i ion, Ha a d School o Public Heal h,
Bos on, MA 02115, USA.
136
Ins i u e o Rep oduc i e and De elopmen al Biology, Impe ial College London, Hamme smi h Hospi al, London W12
0HS, UK.
137
School o Medicine, Uni e si y o Minneso a, Minneapolis, MN 55455, USA.
138
Biocen e Oulu, Uni e si y o Oulu, P.O.
Box 5000Aapis ie 5A, FI-90014 Oulu, Finland.
139
Uni o P ima y Ca e, Oulu Uni e si y Hospi al, Kajaanin ie 50P.O.Box 20FI-90220 Oulu, Finland.
140
Depa men o Child en and Young People and Families, Na ional Ins i u e o Heal h and Wel a e, Aapis ie 1, Box 310, FI-90101 Oulu, Finland.
141
In e uni e si y Ca diology Ins i u e o he Ne he lands, U ech 3511 EP, Ne he lands.
142
Depa men o Ca dio ascula Sciences, Uni e si y o
Leices e , Glenfield Hospi al, Leices e LE3 9QP, UK.
143
Na ional Ins i u e o Heal h Resea ch Leices e Ca dio ascula Biomedical Resea ch Uni ,
Glenfield Hospi al, Leices e LE3 9QP, UK.
144
Ins i u e o Molecula Gene icsCNR, 27100 Pa ia, I aly.
145
Clinical Ins i u e o Medical and Chemical
Labo a o y Diagnos ics, Medical Uni e si y o G az, Auenb ugge pla z 15, 8036 G az, Aus ia.
146
Synlab Academy, Synlab Se ices GmbH, Go lieb-
Daimle -S aße 25, 68165 Mannheim, Ge many.
147
Depa men o Endoc inology, Uni e si y Medical Cen e G oningen, Uni e si y o G oningen,
Hanzeplein 1, 9713 GZ G oningen, Ne he lands.
148
Na ional Hellenic Resea ch Founda ion, Ins i u e o Biological Resea ch and Bio echnology,
A hens 116 35, G eece.
149
Depa men o Toxicogenomics, Maas ich Uni e si y, Uni e si ei ssingel 50, 6229ER Maas ich , Ne he lands.
150
Ox o d Cen e o Diabe es Endoc inology and Me abolism, Uni e si y o Ox o d, Ox o d OX3 7LE, UK
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07345-0
16 NATURE COMMUNICATIONS | (2018) 9:5052 | DOI: 10.1038/s41467-018-07345-0 | www.na u e.c om/na u ecommunica ions