Power of Inclusion: enhancing polygenic prediction with admixed individuals

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

Phenotype: Basophil percentage

Link to the source data field in UK Biobank
Estimated h² in white British population in UKB: 0.034 (95% CI:[0.025, 0.044]).
Link to PGS catalog (Study ID: PGP000502).
GWAS Catalog for mapped trait(s): basophil percentage of leukocytes (EFO_0007992)

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/INI30220/INI30220.WB.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30220/INI30220.NBW.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30220/INI30220.SA.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30220/INI30220.Afr.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30220/INI30220.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	0.002	[0.002, 0.003]	2.0x10^-34
white British	Genotype-only model	R²	0.016	[0.014, 0.018]	5.2x10^-229
white British	Full model (covariates and genotypes)	R²	0.018	[0.016, 0.020]	9.2x10^-258
Non-British white	Covariate-only model	R²	0.004	[-0.001, 0.008]	1.1x10^-03
Non-British white	Genotype-only model	R²	0.018	[0.008, 0.027]	1.7x10^-12
Non-British white	Full model (covariates and genotypes)	R²	0.021	[0.011, 0.031]	1.3x10^-14
South Asian	Covariate-only model	R²	0.000	[-0.002, 0.002]	4.4x10^-01
South Asian	Genotype-only model	R²	0.005	[-0.002, 0.012]	9.4x10^-03
South Asian	Full model (covariates and genotypes)	R²	0.005	[-0.002, 0.012]	1.0x10^-02
African	Covariate-only model	R²	0.002	[-0.003, 0.008]	9.2x10^-02
African	Genotype-only model	R²	0.002	[-0.003, 0.007]	1.3x10^-01
African	Full model (covariates and genotypes)	R²	0.004	[-0.003, 0.012]	2.5x10^-02
Others	Covariate-only model	R²	0.011	[0.007, 0.016]	9.3x10^-21
Others	Genotype-only model	R²	0.012	[0.008, 0.017]	6.4x10^-23
Others	Full model (covariates and genotypes)	R²	0.023	[0.016, 0.029]	7.6x10^-41

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/INI30220/INI30220.BETAs.png

We show the coefficients (BETA) of PGS models. Our iPGS model selected 3472 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

11	72946279	11:72946279:T:C	rs74472890	C	PAVs	P2RY2	0.043
1	236107241	1:236107241:A:C	rs6429432	C	Others	RP5-940F7.2	0.041
19	38901633	19:38901633:C:T	rs34377632	T	PAVs	RASGRP4	0.022
7	92408370	7:92408370:C:T	rs445	T	Intronic	CDK6	-0.021
8	6825295	8:6825295:G:A	rs6996047	A	Others	DEFA10P	0.017
6	82463376	6:82463376:C:T	rs915125	T	Others	FAM46A	-0.014
3	128316435	3:128316435:A:G	rs4328821	G	Others		-0.013
15	91009484	15:91009484:G:A	rs2074585	A	PAVs	IQGAP1	-0.013
3	46250652	3:46250652:C:T	rs3181077	T	Intronic	CCR3	-0.012
10	990143	10:990143:C:T	rs61833265	T	Others	RP11-363N22.2	0.012
19	838178	19:838178:C:G	rs62132293	G	Others	PRTN3	-0.011
X	147554909	X:147554909:G:A	rs16994583	A	Others		0.011
7	92243672	7:92243672:G:A	rs42377	A	UTR	CDK6	0.010
7	143084574	7:143084574:G:A	rs36107836	A	Intronic	ZYX	0.010
19	33754548	19:33754548:C:T	rs78744187	T	Others		-0.010
11	72946140	11:72946140:G:C	rs3741156	C	PAVs	P2RY2	0.010
1	159175354	1:159175354:G:A	rs12075	A	PAVs	DARC	0.009
1	205083865	1:205083865:T:C	rs11240360	C	Intronic	RBBP5	-0.009
19	18285944	19:18285944:G:A	rs11554159	A	PAVs	IFI30	-0.009
15	90969335	15:90969335:G:T	rs2256388	T	PAVs	IQGAP1	-0.009
X	147658782	X:147658782:A:C	rs5936216	C	Intronic	AFF2	-0.009
1	212434241	1:212434241:A:G	rs903119	G	Others		-0.009
8	6904195	8:6904195:A:C	rs6998687	C	Others		0.009
17	56355397	17:56355397:G:A	rs28730837	A	PAVs	MPO	0.008
16	1833508	16:1833508:G:A	rs72761177	A	PTVs	NUBP2	0.008
8	6713695	8:6713695:C:T	rs2980948	T	Intronic	GS1-24F4.2	0.008
19	831815	19:831815:GT:G	rs372854624	G	PTVs	AZU1	0.007
7	75268785	7:75268785:A:G	rs13226566	G	Intronic	HIP1	-0.007
8	6888134	8:6888134:T:C	rs55851618	C	Others	DEFA11P	-0.007
14	88407888	14:88407888:A:G	rs398607	G	PAVs	GALC	0.007
10	26741149	10:26741149:G:T	rs2992261	T	Intronic	APBB1IP	-0.007
1	161479745	1:161479745:A:G	rs1801274	G	PAVs	FCGR2A	0.006
3	128391789	3:128391789:T:C	rs6806687	C	Intronic	RPN1	0.006
6	135498201	6:135498201:C:T	rs1320960	T	Others	MYB	0.006
7	148857795	7:148857795:T:C	rs13225884	C	Intronic	ZNF398	0.006
19	8569735	19:8569735:A:C	rs2967603	C	Others	PRAM1	-0.006
4	87911691	4:87911691:G:T	rs17012234	T	Intronic	AFF1	-0.006
5	126142260	5:126142260:C:T	rs11748362	T	Intronic	LMNB1	0.005
19	38912764	19:38912764:A:G	rs892055	G	PAVs	RASGRP4	0.005
8	6690276	8:6690276:T:C	rs2741098	C	Others	XKR5	0.005
20	43532388	20:43532388:A:G	rs6031860	G	Intronic	YWHAB	-0.005
10	910983	10:910983:C:T	rs10508208	T	Intronic	LARP4B	0.005
8	22974450	8:22974450:T:C	rs9644063	C	PAVs	TNFRSF10C	-0.005
1	118155620	1:118155620:G:A	rs3767812	A	Intronic	FAM46C	-0.005
10	64927823	10:64927823:C:G	rs1935	G	PAVs	JMJD1C	0.005
11	72945341	11:72945341:C:T	rs2511241	T	PAVs	P2RY2	0.005
15	50240200	15:50240200:C:A	rs2413983	A	Intronic	ATP8B4	-0.005
1	235827443	1:235827443:T:C	rs3819013	C	Intronic	LYST	-0.005
4	74948054	4:74948054:A:C	rs1371794	C	Others	RP11-629B11.4	0.004
6	13311851	6:13311851:C:T	rs2496143	T	Intronic	TBC1D7	-0.004
3	187409112	3:187409112:A:C	rs955305	C	Others		0.004
3	128218410	3:128218410:G:A	rs7629791	A	Intronic	RP11-475N22.4	0.004
12	57870155	12:57870155:A:C	rs11544238	C	PAVs	ARHGAP9	0.004
5	150901261	5:150901261:C:T	rs6650971	T	PAVs	FAT2	0.004
22	38205989	22:38205989:T:C	rs2285178	C	PAVs	GCAT	0.004
1	21395357	1:21395357:T:C	rs10916930	C	Intronic	EIF4G3	0.004
19	4409756	19:4409756:A:G	rs2230636	G	PAVs	CHAF1A	0.004
8	6909441	8:6909441:C:T	rs10867025	T	Others	DEFA5	0.004
14	75810616	14:75810616:G:A	rs11621420	A	Others		-0.004
7	50418506	7:50418506:A:C	rs7779749	C	Intronic	IKZF1	0.004
8	126516988	8:126516988:T:C	rs4512391	C	Intronic	RP11-136O12.2	-0.004
8	6789682	8:6789682:A:G	rs2738108	G	Others	DEFA4, GS1-24F4.3	0.004
3	46988561	3:46988561:C:T	rs13092573	T	Intronic	CCDC12	0.004
2	27260469	2:27260469:G:A	rs1124649	A	PAVs	TMEM214	-0.004
17	37814080	17:37814080:G:A	rs1877031	A	PAVs	STARD3	-0.004
19	795816	19:795816:G:A	rs8113356	A	Others	PTBP1	0.004
1	17674537	1:17674537:C:A	rs2240335	A	PCVs	PADI4	0.004
5	126161266	5:126161266:A:C	rs3828699	C	Intronic	LMNB1	0.004
9	130561688	9:130561688:T:C	rs12379987	C	Intronic	FPGS	-0.004
8	61660163	8:61660163:A:G	rs11775560	G	Intronic	CHD7	-0.004
21	39527971	21:39527971:A:G	rs991015	G	Intronic	DSCR8	0.004
3	50332697	3:50332697:G:A	rs13100173	A	PAVs	HYAL3	-0.004
7	5571625	7:5571625:C:T	rs2966449	T	Intronic	ACTB	-0.004
4	103418957	4:103418957:T:C	rs980455	C	Others	NFKB1	-0.004
7	50417632	7:50417632:A:G	rs62447197	G	Intronic	IKZF1	0.004
2	127587337	2:127587337:T:C	rs10197495	C	Others		-0.004
14	23589057	14:23589057:G:A	rs2239633	A	Others	CEBPE	0.004
20	30385192	20:30385192:C:T	rs6058463	T	PAVs	TPX2	-0.004
1	153336018	1:153336018:C:G	rs724781	G	Others	S100A9	-0.004
1	205084120	1:205084120:G:A	rs7515178	A	Intronic	RBBP5	-0.004
17	38175553	17:38175553:C:T	rs709592	T	UTR	MED24	0.004
8	6698224	8:6698224:A:G	rs3863728	G	Intronic	GS1-24F4.2	0.004
10	102104521	10:102104521:C:T	rs735877	T	Intronic	RP11-34D15.2	-0.004
10	99116903	10:99116903:C:T	rs1048445	T	PAVs	RRP12	-0.004
19	18550276	19:18550276:C:A	rs10427083	A	Others	CTD-3137H5.1, ISYNA1	0.003
19	862355	19:862355:G:C	rs62132300	C	Intronic	CFD	0.003
3	128154858	3:128154858:A:G	rs1876284	G	Others		-0.003
3	16894193	3:16894193:T:C	rs2347655	C	Intronic	PLCL2	-0.003
20	44974696	20:44974696:T:G	rs6017787	G	Others	SLC35C2	-0.003
19	49244218	19:49244218:C:CAA	rs1188333414	CAA	PTVs	IZUMO1	-0.003
3	73111998	3:73111998:C:T	rs1060584	T	PAVs	EBLN2	0.003
17	59940633	17:59940633:C:T	rs4988340	T	Intronic	BRIP1	0.003
11	118967758	11:118967758:T:C	rs643788	C	PAVs	DPAGT1	0.003
17	56492800	17:56492800:T:C	rs3744093	C	PAVs	RNF43	0.003
1	212431431	1:212431431:A:G	rs7527863	G	Others		-0.003
1	224624391	1:224624391:T:C	rs6426123	C	Intronic	WDR26, CNIH3	0.003
2	65608909	2:65608909:C:T	rs1876518	T	Intronic	SPRED2	0.003
11	68855363	11:68855363:G:A	rs3829241	A	PAVs	TPCN2	0.003
7	50300869	7:50300869:G:A	rs10248090	A	Others		0.003
2	148692776	2:148692776:T:C	rs3768686	C	UTR	ORC4	-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 3472 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.