Power of Inclusion: enhancing polygenic prediction with admixed individuals

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

Phenotype: Monocyte count

Link to the source data field in UK Biobank
Estimated h² in white British population in UKB: 0.098 (95% CI:[0.084, 0.112]).
Link to PGS catalog (Study ID: PGP000502).
GWAS Catalog for mapped trait(s): monocyte count (EFO_0005091)

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/INI30130/INI30130.WB.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30130/INI30130.NBW.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30130/INI30130.SA.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30130/INI30130.Afr.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30130/INI30130.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	0.042	[0.039, 0.045]	<1.0x10^-300
white British	Genotype-only model	R²	0.082	[0.078, 0.086]	<1.0x10^-300
white British	Full model (covariates and genotypes)	R²	0.124	[0.120, 0.129]	<1.0x10^-300
Non-British white	Covariate-only model	R²	0.057	[0.040, 0.073]	2.1x10^-37
Non-British white	Genotype-only model	R²	0.090	[0.070, 0.110]	2.1x10^-59
Non-British white	Full model (covariates and genotypes)	R²	0.145	[0.121, 0.169]	1.2x10^-97
South Asian	Covariate-only model	R²	0.044	[0.024, 0.065]	8.6x10^-16
South Asian	Genotype-only model	R²	0.070	[0.045, 0.095]	2.2x10^-24
South Asian	Full model (covariates and genotypes)	R²	0.113	[0.082, 0.143]	7.8x10^-39
African	Covariate-only model	R²	0.005	[-0.003, 0.013]	1.3x10^-02
African	Genotype-only model	R²	0.033	[0.014, 0.053]	3.9x10^-10
African	Full model (covariates and genotypes)	R²	0.033	[0.014, 0.053]	4.2x10^-10
Others	Covariate-only model	R²	0.043	[0.034, 0.051]	3.0x10^-75
Others	Genotype-only model	R²	0.079	[0.068, 0.091]	6.1x10^-141
Others	Full model (covariates and genotypes)	R²	0.111	[0.098, 0.124]	6.0x10^-199

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/INI30130/INI30130.BETAs.png

We show the coefficients (BETA) of PGS models. Our iPGS model selected 13415 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

22	17586757	22:17586757:T:C	rs140221307	C	PAVs	IL17RA	-0.056
13	28623048	13:28623048:T:C	rs79490353	C	Intronic	FLT3	0.037
1	92554283	1:92554283:G:A	rs34856868	A	PAVs	BTBD8	0.021
3	42906116	3:42906116:T:C	rs2228467	C	PAVs	ACKR2	0.018
2	182319301	2:182319301:C:T	rs1449263	T	Others	ITGA4	0.018
12	69744014	12:69744014:C:A	rs1800973	A	PAVs	LYZ	0.017
17	16852187	17:16852187:A:G	rs34557412	G	PAVs	TNFRSF13B	-0.012
7	92408370	7:92408370:C:T	rs445	T	Intronic	CDK6	-0.010
9	113915905	9:113915905:T:C	rs10980800	C	Intronic	RP11-202G18.1	0.010
1	236107241	1:236107241:A:C	rs6429432	C	Others	RP5-940F7.2	-0.009
8	130597585	8:130597585:C:A	rs2163950	A	Intronic	CCDC26	-0.009
3	39307162	3:39307162:G:A	rs3732378	A	PAVs	CX3CR1	0.009
16	85938835	16:85938835:G:A	rs11646550	A	Intronic	IRF8	-0.008
16	85966548	16:85966548:A:G	rs76121846	G	Others	RP11-542M13.3	0.008
13	28624294	13:28624294:G:A	rs1933437	A	PAVs	FLT3	0.008
15	80263217	15:80263217:C:T	rs3826007	T	PAVs	BCL2A1	-0.007
1	159175494	1:159175494:C:T	rs34599082	T	PAVs	DARC	-0.007
7	50417632	7:50417632:A:G	rs62447197	G	Intronic	IKZF1	-0.007
14	103847274	14:103847274:C:T	rs8020912	T	Others	MARK3	0.007
19	54327313	19:54327313:C:A	rs34436714	A	PAVs	NLRP12	-0.007
15	64634884	15:64634884:C:A	rs6494475	A	PAVs	CTD-2116N17.1	0.007
9	113919469	9:113919469:G:T	rs10217127	T	Intronic	RP11-202G18.1	-0.007
19	45740771	19:45740771:C:T	rs17356664	T	Intronic	MARK4	-0.007
12	111884608	12:111884608:T:C	rs3184504	C	PAVs	SH2B3	-0.006
16	85933038	16:85933038:C:T	rs2270502	T	Intronic	IRF8	-0.006
20	31191015	20:31191015:C:T	rs12480732	T	Intronic	RP11-410N8.4	0.006
12	6500843	12:6500843:G:A	rs4301834	A	Others	RP1-102E24.8	-0.006
5	149498151	5:149498151:G:A	rs246394	A	Intronic	PDGFRB	-0.006
3	128303373	3:128303373:C:T	rs6798431	T	Others		0.006
20	48956954	20:48956954:A:G	rs4811031	G	Others		-0.006
16	85995436	16:85995436:T:C	rs113646461	C	Others		0.006
7	50305863	7:50305863:T:G	rs4917014	G	Others		-0.005
8	130577382	8:130577382:T:G	rs9649961	G	Intronic	CCDC26	-0.005
17	57968398	17:57968398:AC:A	rs143803908	A	PTVs	TUBD1	-0.005
3	46399208	3:46399208:G:A	rs1799864	A	PAVs	CCR2	-0.005
8	130685457	8:130685457:T:G	rs4295627	G	Intronic	CCDC26	-0.005
14	23589057	14:23589057:G:A	rs2239633	A	Others	CEBPE	0.005
18	60920854	18:60920854:C:T	rs17758695	T	Intronic	BCL2	-0.005
6	135419018	6:135419018:T:C	rs9399137	C	Intronic	HBS1L	-0.005
9	113911613	9:113911613:T:C	rs7870066	C	Intronic	RP11-202G18.1	-0.005
7	6502367	7:6502367:T:C	rs6796	C	UTR	KDELR2	0.005
16	85917551	16:85917551:T:C	rs9937847	C	Others		-0.005
20	48909280	20:48909280:C:T	rs17196808	T	Others	RP11-290F20.1	-0.004
19	49979398	19:49979398:G:A	rs17272847	A	PAVs	FLT3LG	0.004
14	25459482	14:25459482:T:C	rs2332462	C	Intronic	STXBP6	0.004
9	91460248	9:91460248:C:T	rs2174057	T	Others		-0.004
15	42261781	15:42261781:G:A	rs1002774	A	Intronic	EHD4	-0.004
10	25218243	10:25218243:G:T	rs10828725	T	Intronic	PRTFDC1	-0.004
20	48894424	20:48894424:A:C	rs2274950	C	Others	RP11-290F20.3	0.004
16	86009760	16:86009760:C:A	rs12232384	A	Others		0.004
1	150902941	1:150902941:A:G	rs11204744	G	Intronic	SETDB1	-0.004
3	46306700	3:46306700:T:C	rs4987053	C	PCVs	CCR3	-0.004
13	41003022	13:41003022:C:T	rs9315776	T	Others		-0.004
6	41991061	6:41991061:A:G	rs4714552	G	Intronic	CCND3	-0.004
4	83570300	4:83570300:T:C	rs17005996	C	Intronic	SCD5	-0.004
9	113948397	9:113948397:A:G	rs17207110	G	Intronic	RP11-202G18.1	-0.004
1	150571730	1:150571730:A:G	rs2048558	G	Others	ENSA, SNORA40	-0.004
9	113945046	9:113945046:A:G	rs12339649	G	Intronic	RP11-202G18.1	0.004
19	18389135	19:18389135:A:G	rs12608504	G	Others	KIAA1683, MIR3188	0.004
6	82463376	6:82463376:C:T	rs915125	T	Others	FAM46A	0.004
15	80263406	15:80263406:C:T	rs1138357	T	PAVs	BCL2A1	-0.004
7	50258313	7:50258313:C:T	rs1870028	T	Others		-0.004
8	130679176	8:130679176:A:G	rs6985166	G	Intronic	CCDC26	-0.004
9	113958351	9:113958351:C:A	rs2039183	A	Intronic	RP11-202G18.1	0.004
16	85991705	16:85991705:A:C	rs11642873	C	Others		0.004
2	136883823	2:136883823:C:T	rs4954391	T	Others		-0.004
10	101274365	10:101274365:G:A	rs7078219	A	Others		-0.003
7	50473751	7:50473751:G:A	rs6944602	A	Others	IKZF1	0.003
14	93523894	14:93523894:C:T	rs749619	T	Intronic	ITPK1	0.003
2	241569692	2:241569692:C:T	rs3749171	T	PAVs	GPR35	-0.003
10	99072036	10:99072036:T:C	rs11189121	C	Others	RP11-452K12.3	-0.003
11	122499895	11:122499895:A:G	rs11607161	G	Others		0.003
17	81103682	17:81103682:T:C	rs9330525	C	Intronic	AC139099.4	0.003
13	28482654	13:28482654:T:C	rs9581941	C	Intronic	PDX1-AS1	0.003
3	188087628	3:188087628:C:T	rs9851967	T	Intronic	LPP	-0.003
12	9098995	12:9098995:G:GAAC	rs3217106	GAAC	PAVs	M6PR	0.003
19	44281898	19:44281898:G:A	rs11879798	A	Intronic	KCNN4	-0.003
6	31506691	6:31506691:G:A	rs2071596	A	PAVs	DDX39B	-0.003
8	130627555	8:130627555:T:C	rs7814618	C	Intronic	CCDC26	-0.003
2	111789791	2:111789791:A:G	rs13012948	G	Intronic	ACOXL	-0.003
12	108724296	12:108724296:G:T	rs952584	T	Intronic	CMKLR1	-0.003
20	8604593	20:8604593:G:A	rs6077396	A	Intronic	PLCB1	-0.003
9	22145694	9:22145694:A:G	rs2065500	G	Others		-0.003
17	57875396	17:57875396:C:T	rs2665404	T	Intronic	VMP1	0.003
13	40984601	13:40984601:A:G	rs2324606	G	Others	RN7SKP2	-0.003
11	128546634	11:128546634:G:T	rs580481	T	Intronic	RP11-744N12.3	0.003
13	108979043	13:108979043:T:C	rs1224174	C	Others		0.003
22	17589209	22:17589209:C:T	rs879577	T	PAVs	IL17RA	0.003
16	85927814	16:85927814:C:T	rs391023	T	Others	IRF8	-0.003
6	7167170	6:7167170:A:G	rs9379077	G	Intronic	RREB1	0.003
15	80194247	15:80194247:C:T	rs2115536	T	Intronic	ST20, ST20-MTHFS	0.003
2	111931421	2:111931421:T:C	rs726430	C	Others		-0.003
5	149481778	5:149481778:C:T	rs7722898	T	Intronic	CSF1R	0.003
13	41216534	13:41216534:C:T	rs4325427	T	Intronic	FOXO1	-0.003
5	149436545	5:149436545:G:A	rs216137	A	Intronic	CSF1R	0.003
2	225727663	2:225727663:C:T	rs6436515	T	Intronic	DOCK10	-0.003
3	12267648	3:12267648:A:G	rs7616006	G	Others		-0.003
6	32411035	6:32411035:A:C	rs8084	C	PTVs	HLA-DRA	0.003
1	243798208	1:243798208:G:T	rs320330	T	Intronic	AKT3	0.003
17	25843643	17:25843643:G:A	rs2945412	A	Intronic	KSR1	-0.003

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 13415 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.