Power of Inclusion: enhancing polygenic prediction with admixed individuals

Tanigawa and Kellis. Am J Hum Genet. (2023).

Phenotype: Hematocrit percentage

Link to the source data field in UK Biobank
Estimated h² in white British population in UKB: 0.140 (95% CI:[0.125, 0.155]).
Link to PGS catalog (Study ID: PGP000502).
GWAS Catalog for mapped trait(s): hematocrit (EFO_0004348)

Our FAQ page shows the description of the file format and how you may use iPGS coefficients in your research.

iPGS prediction in the held-out test set individuals

We compared the polygenic prediction from our iPGS model and the phenotype values using the held-out test set individuals in UK Biobank. Note the difference in the number of individuals in the five population groups.

/static/data/tanigawakellis2023/per_trait/INI30030/INI30030.WB.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30030/INI30030.NBW.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30030/INI30030.SA.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30030/INI30030.Afr.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30030/INI30030.others.PGS_vs_phe.png

Predictive performance

Population	Model	Metric	Predictive Performance	95% CI	P-value
Population	Model	Metric	Predictive Performance	95% CI	P-value

white British	Covariate-only model	R²	0.333	[0.327, 0.339]	<1.0x10^-300
white British	Genotype-only model	R²	0.067	[0.064, 0.071]	<1.0x10^-300
white British	Full model (covariates and genotypes)	R²	0.400	[0.394, 0.405]	<1.0x10^-300
Non-British white	Covariate-only model	R²	0.367	[0.339, 0.395]	1.3x10^-281
Non-British white	Genotype-only model	R²	0.058	[0.041, 0.074]	2.3x10^-38
Non-British white	Full model (covariates and genotypes)	R²	0.426	[0.398, 0.453]	<1.0x10^-300
South Asian	Covariate-only model	R²	0.390	[0.351, 0.429]	1.4x10^-155
South Asian	Genotype-only model	R²	0.033	[0.015, 0.051]	3.9x10^-12
South Asian	Full model (covariates and genotypes)	R²	0.426	[0.388, 0.464]	2.5x10^-174
African	Covariate-only model	R²	0.406	[0.363, 0.448]	2.3x10^-132
African	Genotype-only model	R²	0.016	[0.002, 0.030]	1.5x10^-05
African	Full model (covariates and genotypes)	R²	0.413	[0.370, 0.455]	2.4x10^-135
Others	Covariate-only model	R²	0.359	[0.342, 0.375]	<1.0x10^-300
Others	Genotype-only model	R²	0.049	[0.039, 0.058]	6.3x10^-86
Others	Full model (covariates and genotypes)	R²	0.404	[0.387, 0.420]	<1.0x10^-300

The predictive performance (R²), its 95% confidence interval (CI), and statistical significance (P-value) are shown for each population in UK Biobank in the held-out test set. The "model" column indicates whether the predictive performance is from the covariate-terms alone (covariate-only model), PGS terms alone (Genotype-only model), or the full model containing both PGS and covariate terms. We used the following sets of covariates in our analysis: age, sex, age², age*sex, Townsend deprivation index, and genotype PCs (PC1-PC18). Please refer to our publication for a more detailed description of the methods.

Coefficients (BETA) of PGS models

/static/data/tanigawakellis2023/per_trait/INI30030/INI30030.BETAs.png

We show the coefficients (BETA) of PGS models. Our iPGS model selected 20106 variants with non-zero coefficients. The genetic variants with the large absolute values of coefficients are annotated in the plot. There is no guarantee that our iPGS model selects causal variants. We use the GRCh37/hg19 reference genome.

The top 100 genetic variants with the largest absolute value of coefficients

CHROM	POS	Variant	Variant ID	Effect allele	Consequence	Gene symbol	Beta
CHROM	POS	Variant	Variant ID	Effect allele	Consequence	Gene symbol	Beta

7	100275444	7:100275444:G:A	rs62482253	A	PAVs	GNB2	-0.284
10	71093392	10:71093392:C:T	rs16926246	T	Intronic	HK1	0.212
6	26091179	6:26091179:C:G	rs1799945	G	PAVs	HFE	0.188
6	135419018	6:135419018:T:C	rs9399137	C	Intronic	HBS1L	-0.186
7	151415041	7:151415041:A:G	rs10224002	G	Intronic	PRKAG2	-0.158
22	37469590	22:37469590:C:T	rs387907018	T	PAVs	TMPRSS6	0.156
6	25918225	6:25918225:T:C	rs80215559	C	Intronic	SLC17A2	0.150
12	111884608	12:111884608:T:C	rs3184504	C	PAVs	SH2B3	-0.145
2	46353166	2:46353166:A:G	rs10495928	G	Intronic	PRKCE	-0.141
19	41306650	19:41306650:C:T	rs61750953	T	PAVs	EGLN2	-0.115
15	43820717	15:43820717:C:T	rs55707100	T	PAVs	MAP1A	-0.109
9	136131022	9:136131022:C:T	rs8176751	T	Others	ABO	0.106
10	71099913	10:71099913:T:C	rs7072268	C	Intronic	HK1	0.104
4	103188709	4:103188709:C:T	rs13107325	T	PAVs	SLC39A8	0.104
6	43941137	6:43941137:T:C	rs17287978	C	Others		-0.103
9	136153875	9:136153875:C:T	rs651007	T	Others	ABO	-0.099
1	155178782	1:155178782:A:T	rs760077	T	PAVs	MTX1	0.096
1	214180118	1:214180118:G:A	rs726334	A	Intronic	PROX1	0.095
10	71112843	10:71112843:G:T	rs10998738	T	Intronic	HK1	-0.091
12	48512285	12:48512285:C:A	rs4760682	A	PAVs	PFKM	-0.090
11	30754837	11:30754837:G:A	rs55733296	A	Others		-0.082
6	25842951	6:25842951:T:G	rs1408272	G	Intronic	SLC17A3	0.079
20	43042364	20:43042364:C:T	rs1800961	T	PAVs	HNF4A	0.078
1	154965113	1:154965113:G:T	rs7535144	T	PAVs	FLAD1	-0.077
15	76298132	15:76298132:A:G	rs4886755	G	PCVs	NRG4	-0.076
22	46364161	22:46364161:G:A	rs9330813	A	Intronic	WNT7B	0.074
3	56771251	3:56771251:A:C	rs3772219	C	PAVs	ARHGEF3	-0.074
4	55394172	4:55394172:C:T	rs218237	T	Others		-0.073
16	88853729	16:88853729:C:T	rs837763	T	Others	PIEZO1	-0.071
9	130622946	9:130622946:T:C	rs4837197	C	Others		-0.071
17	59456589	17:59456589:C:T	rs9895661	T	Others	BCAS3	-0.070
2	46315516	2:46315516:C:G	rs71422190	G	Intronic	PRKCE	-0.068
19	33754548	19:33754548:C:T	rs78744187	T	Others		0.067
2	120848049	2:120848049:C:T	rs28930677	T	PAVs	EPB41L5	0.066
15	76232982	15:76232982:C:T	rs2648437	T	Intronic	NRG4	-0.066
6	43801582	6:43801582:C:T	rs12660375	T	Others		-0.065
7	100361675	7:100361675:G:A	rs2293767	A	PAVs	ZAN	0.064
2	113972945	2:113972945:A:G	rs752590	G	Intronic	PAX8-AS1	0.063
19	807442	19:807442:G:C	rs123698	C	Others	PTBP1	0.062
X	8916646	X:8916646:A:C	rs17307280	C	Others		0.062
19	41328365	19:41328365:C:T	rs11879672	T	Intronic	CYP2F2P, CTC-490E21.12	-0.062
19	41381683	19:41381683:G:T	rs58682606	T	PTVs	CYP2A7	-0.060
6	43806609	6:43806609:G:A	rs881858	A	Others		0.059
1	147282508	1:147282508:C:T	rs11240129	T	Others		0.058
14	64703593	14:64703593:G:T	rs1256061	T	Intronic	ESR2	-0.057
1	231488524	1:231488524:C:T	rs2437150	T	PAVs	SPRTN	0.057
2	46359079	2:46359079:A:G	rs17034610	G	Intronic	PRKCE	-0.055
7	100315517	7:100315517:A:G	rs1734907	G	Others	EPO	-0.055
19	19379549	19:19379549:C:T	rs58542926	T	PAVs	TM6SF2	0.054
1	46493460	1:46493460:T:G	rs1707336	G	PAVs	MAST2	-0.054
1	48030350	1:48030350:G:A	rs11583750	A	Others		-0.053
21	16339172	21:16339172:G:C	rs2229742	C	PAVs	NRIP1	-0.053
4	77410318	4:77410318:C:A	rs4859682	A	Intronic	SHROOM3	0.053
7	1274582	7:1274582:G:A	rs4724799	A	Intronic	UNCX	-0.051
20	57597970	20:57597970:A:C	rs463312	C	PAVs	TUBB1	0.051
22	44324727	22:44324727:C:G	rs738409	G	PAVs	PNPLA3	0.051
15	58723426	15:58723426:A:G	rs1077835	G	Intronic	LIPC, ALDH1A2	-0.051
9	6356657	9:6356657:C:T	rs7021445	T	Others		0.050
21	35358762	21:35358762:G:A	rs2834322	A	Others		-0.050
11	30749090	11:30749090:T:C	rs963837	C	Others		0.049
16	30103160	16:30103160:C:A	rs3809627	A	UTR	TBX6	0.049
2	121991967	2:121991967:C:T	rs11677172	T	Intronic	TFCP2L1	0.048
2	44017188	2:44017188:T:C	rs3792020	C	Intronic	DYNC2LI1	0.048
17	43924073	17:43924073:T:C	rs12373123	C	PAVs	SPPL2C	0.048
12	13046211	12:13046211:C:T	rs12811512	T	Intronic	GPRC5A	-0.048
2	46306040	2:46306040:G:A	rs113797384	A	Intronic	PRKCE	-0.048
19	2170764	19:2170764:T:C	rs2108524	C	Intronic	DOT1L	0.047
1	10424252	1:10424252:C:T	rs11121548	T	Intronic	KIF1B	0.047
20	61038322	20:61038322:C:T	rs6121609	T	Others	GATA5	0.047
7	150930363	7:150930363:C:T	rs73169668	T	UTR	CHPF2	-0.046
6	43736389	6:43736389:A:C	rs699947	C	Others	VEGFA	-0.046
3	69870595	3:69870595:A:G	rs1529585	G	Intronic	MITF	0.046
13	29230581	13:29230581:A:G	rs1340817	G	Others	POMP	-0.046
1	29320013	1:29320013:G:A	rs111642750	A	PAVs	EPB41	0.045
11	10174447	11:10174447:C:T	rs7129531	T	Intronic	RP11-748C4.1, SBF2	-0.045
4	23736523	4:23736523:A:G	rs16874052	G	Intronic	RP11-380P13.1	0.044
17	4835895	17:4835895:T:C	rs2243093	C	PAVs	GP1BA	0.043
16	72144174	16:72144174:T:C	rs9302635	C	Intronic	DHX38	-0.043
16	86339661	16:86339661:C:T	rs11649143	T	Others		-0.043
6	31106501	6:31106501:C:CC	Affx-89026413	CC	PTVs	PSORS1C1	-0.043
15	33324519	15:33324519:T:C	rs17816699	C	Intronic	FMN1	0.043
8	145112983	8:145112983:C:T	rs55916375	T	PAVs	OPLAH	0.042
6	12295876	6:12295876:A:G	rs1629862	G	Intronic	EDN1	-0.042
3	12359049	3:12359049:G:A	rs2067819	A	Intronic	PPARG	-0.042
12	2518352	12:2518352:T:C	rs4765929	C	Intronic	CACNA1C	0.042
14	36158828	14:36158828:T:C	rs7155504	C	Others	RALGAPA1	-0.042
2	177042633	2:177042633:A:C	rs2072590	C	Others	HOXD-AS1, AC009336.24	0.042
12	57843711	12:57843711:G:A	rs2229357	A	PAVs	INHBC	0.041
8	116559435	8:116559435:A:G	rs3808434	G	Intronic	TRPS1	0.040
13	114549015	13:114549015:T:C	rs6602910	C	Intronic	GAS6	0.040
7	129663496	7:129663496:C:T	rs11556924	T	PAVs	ZC3HC1	-0.040
16	16259596	16:16259596:G:A	rs41278174	A	PAVs	ABCC6	0.040
1	28564279	1:28564279:A:G	rs9508	G	PAVs	ATPIF1	0.039
8	9185146	8:9185146:T:C	rs2126259	C	Intronic	RP11-115J16.1	-0.039
3	58410554	3:58410554:C:T	rs34579268	T	PAVs	PXK	-0.039
20	25282967	20:25282967:C:T	rs746748	T	PAVs	ABHD12	-0.039
2	46372675	2:46372675:A:G	rs10495930	G	Intronic	PRKCE	-0.039
4	88052219	4:88052219:T:C	rs342467	C	PAVs	AFF1	-0.039
6	31238217	HLA-C*0501	HLA-C*0501	+	PAVs	HLA-C	-0.039
22	38877461	22:38877461:T:G	rs12004	G	PAVs	KDELR3	-0.039

There is no guarantee that our iPGS model selects causal variants. We show the top 100 variants with the largest effect size (BETA). To see 20106 variants included in our iPGS model, please download the iPGS coefficients by clicking the download button. We use the GRCh37/hg19 reference genome.

Follow-up analysis

There are several ways to use the resource in your research. First, you may use our iPGS coefficients and compute individual-level polygenic scores for your cohort. Second, you may also investigate the genetic variants with non-zero coefficients and their annotated genes to learn more about biology by taking advantage of the sparsity of our iPGS models. For your convenience, here we suggest several resources as an example of follow-up analysis. We do not intend to cover all the relevant follow-up analyses.

Using iPGS coefficients

By clicking the download button above, you may download the iPGS coefficients. Our FAQ page shows the description of file format and how you may use iPGS coefficients in your research.

HaploReg

HaploReg is a tool for exploring annotations of the non-coding genome at variants on haplotype blocks. The button above submits the top 100 genetic variants with the largest absolute value of coefficients as a query to HaploReg using the default parameters in HaploReg v4.2 (LD threshold r² >= 1, ChromHMM 15-state model, SiPhy-omega, and GENCODE genes). HaploReg's ability to browse haplotypes is useful here as there is no guarantee that our iPGS model selects causal variants. The 'top 100 variant' cutoff is an arbitrary threshold; we aim to demonstrate how one may investigate the selected variants. Please check Ward and Kellis. Nucleic Acids Res. 2012 and Ward and Kellis. Nucleic Acids Res. 2016 for more information on HaploReg.

GREAT

GREAT: Genomic Regions Enrichment of Annotations Tool evaluates enrichment of pathway and ontology terms. The ability of GREAT to map non-coding genetic variants to their downstream target genes would be suitable for investigating pathway and ontology enrichment of genetic variants selected in our sparse iPGS model. The button above submits the top 1000 genetic variants with the largest absolute value of coefficients as a query to GREAT using the default parameters in GREAT v4.0.4. The 'top 1000 variant' cutoff is an arbitrary threshold; we aim to demonstrate how one may investigate the selected variants. Please check McLean et al. Nat Biotechnol. 2010 and Tanigawa*, Dyer*, and Bejerano. PLoS Comput Biol. 2022 for more information on GREAT.

References

Y. Tanigawa and M. Kellis. Power of inclusion: Enhancing polygenic prediction with admixed individuals. Am J Hum Genet. (2023).
- We introduce our inclusive PGS approach in this publication.
- Supplementary Data Files at figshare (doi: 10.6084/m9.figshare.22905368)
- MIT News article: Making genetic prediction models more inclusive
- Author interview: Inside AJHG: A Chat with Yosuke Tanigawa
J. Qian, Y. Tanigawa, W. Du, M. Aguirre, C. Chang, R. Tibshirani, M. A. Rivas, T. Hastie. A fast and scalable framework for large-scale and ultrahigh-dimensional sparse regression with application to the UK Biobank. PLoS Genet. 16, e1009141 (2020).
- In iPGS, we fit penalized multivariate regression on individual-level data using the BASIL algorithm implemented in the R snpnet package. This publication describes the BASIL algorithm and the R snpnet package.