Power of Inclusion: enhancing polygenic prediction with admixed individuals

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

Phenotype: High light scatter reticulocyte percentage

Link to the source data field in UK Biobank
Estimated h² in white British population in UKB: 0.068 (95% CI:[0.058, 0.078]).
Link to PGS catalog (Study ID: PGP000502).
GWAS Catalog for mapped trait(s): reticulocyte measurement (EFO_0010700)

High light scatter reticulocyte % iPGS coefficients

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/INI30290/INI30290.WB.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30290/INI30290.NBW.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30290/INI30290.SA.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30290/INI30290.Afr.PGS_vs_phe.png

/static/data/tanigawakellis2023/per_trait/INI30290/INI30290.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	0.001	[0.001, 0.002]	4.9x10^-19
white British	Genotype-only model	R²	0.017	[0.015, 0.019]	4.0x10^-246
white British	Full model (covariates and genotypes)	R²	0.018	[0.016, 0.020]	1.4x10^-262
Non-British white	Covariate-only model	R²	0.006	[0.000, 0.012]	4.3x10^-05
Non-British white	Genotype-only model	R²	0.099	[0.078, 0.119]	1.5x10^-64
Non-British white	Full model (covariates and genotypes)	R²	0.105	[0.083, 0.126]	2.6x10^-68
South Asian	Covariate-only model	R²	0.016	[0.003, 0.029]	2.0x10^-06
South Asian	Genotype-only model	R²	0.088	[0.060, 0.115]	1.5x10^-29
South Asian	Full model (covariates and genotypes)	R²	0.097	[0.069, 0.126]	8.1x10^-33
African	Covariate-only model	R²	0.000	[-0.002, 0.003]	4.7x10^-01
African	Genotype-only model	R²	0.031	[0.012, 0.050]	2.7x10^-09
African	Full model (covariates and genotypes)	R²	0.030	[0.011, 0.049]	5.3x10^-09
Others	Covariate-only model	R²	0.017	[0.012, 0.023]	1.6x10^-30
Others	Genotype-only model	R²	0.083	[0.071, 0.094]	1.7x10^-144
Others	Full model (covariates and genotypes)	R²	0.095	[0.083, 0.107]	5.0x10^-166

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/INI30290/INI30290.BETAs.png

We show the coefficients (BETA) of PGS models. Our iPGS model selected 11641 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

6	139842599	6:139842599:G:T	rs653513	T	Others		-0.027
5	52080909	5:52080909:T:C	rs77704739	C	Intronic	CTD-2288O8.1	0.023
1	248039294	1:248039294:G:A	rs1339847	A	PAVs	TRIM58	0.019
1	158637728	1:158637728:T:C	rs148912436	C	PAVs	SPTA1	0.018
20	4157136	20:4157136:G:A	rs1741317	A	Intronic	SMOX	-0.017
19	11294120	19:11294120:T:C	rs17678527	C	Intronic	KANK2	0.016
6	139840693	6:139840693:A:C	rs592423	C	Others		-0.016
X	153763492	X:153763492:T:C	rs1050829	C	PAVs	G6PD	0.013
8	41543675	8:41543675:G:A	rs34664882	A	PAVs	ANK1	0.013
5	52096889	5:52096889:C:A	rs1499280	A	PAVs	PELO	-0.012
10	25218243	10:25218243:G:T	rs10828725	T	Intronic	PRTFDC1	-0.012
3	50352458	3:50352458:T:G	rs9877046	G	Others	HYAL1	-0.012
20	4172258	20:4172258:C:T	rs13042073	T	Others	RP4-779E11.3	0.011
6	30128442	6:30128442:C:T	rs12212092	T	PAVs	TRIM10	-0.011
8	41630405	8:41630405:G:A	rs4737009	A	Intronic	ANK1	-0.010
20	4155948	20:4155948:G:A	rs1741315	A	Intronic	SMOX	-0.010
20	4169079	20:4169079:G:A	rs16989303	A	Others	RP4-779E11.3	0.010
X	40833508	X:40833508:G:T	rs5963904	T	Others		0.009
12	120900274	12:120900274:C:T	rs9040	T	Others	SRSF9	-0.009
7	6066450	7:6066450:T:C	rs2640	C	PAVs	EIF2AK1	0.009
12	112331317	12:112331317:G:A	rs12580246	A	PTVs	MAPKAPK5	-0.009
1	248039451	1:248039451:C:T	rs3811444	T	PAVs	TRIM58	0.008
5	95163449	5:95163449:G:A	rs6556886	A	Others	GLRX	-0.008
6	30126303	6:30126303:T:A	rs61737427	A	PAVs	TRIM10	-0.008
1	118254209	1:118254209:A:G	rs11580552	G	Others		0.007
1	158577109	1:158577109:A:C	rs857685	C	PAVs	OR10Z1	-0.007
17	33884804	17:33884804:T:C	rs10512472	C	PAVs	SLFN14	0.007
14	23494277	14:23494277:A:G	rs8013143	G	Intronic	PSMB5	-0.007
5	52095943	5:52095943:G:A	rs114309882	A	UTR	PELO	0.007
6	26091179	6:26091179:C:G	rs1799945	G	PAVs	HFE	0.007
10	104637992	10:104637992:A:G	rs10786719	G	Intronic	AS3MT, C10orf32-ASMT	-0.007
7	134389713	7:134389713:C:T	rs6944563	T	Others		-0.007
5	154056002	5:154056002:C:T	rs13186731	T	Others		0.007
6	30121460	6:30121460:C:T	rs2022065	T	UTR	TRIM10	-0.006
12	111884608	12:111884608:T:C	rs3184504	C	PAVs	SH2B3	-0.006
2	220081416	2:220081416:G:A	rs57467915	A	PAVs	ABCB6	0.006
1	56328596	1:56328596:G:T	rs4926698	T	PAVs	RP11-90C4.1	-0.006
1	45810865	1:45810865:G:A	rs17853159	A	PAVs	TESK2	0.006
2	24148231	2:24148231:T:C	rs963725	C	Intronic	ATAD2B	-0.006
3	49689210	3:49689210:G:A	rs34762726	A	PAVs	BSN	-0.006
16	212649	16:212649:C:T	rs3785309	T	Intronic	HBM	-0.006
14	74237247	14:74237247:G:A	rs8009224	A	Intronic	ELMSAN1	0.006
X	70352417	X:70352417:T:C	rs10521349	C	Intronic	MED12	0.005
9	140117968	9:140117968:A:G	rs73565707	G	Others	C9orf169, RNF224, RNF208	0.005
5	95163437	5:95163437:A:G	rs17462893	G	Others	GLRX	0.005
4	55394172	4:55394172:C:T	rs218237	T	Others		0.005
1	3691997	1:3691997:AGTCAGCCTAGGGGCTGT:A	rs566629828	A	PTVs	SMIM1	-0.005
17	33881631	17:33881631:T:C	rs321612	C	PAVs	SLFN14	-0.005
19	10731745	19:10731745:T:C	rs8112355	C	Intronic	SLC44A2	0.005
4	144890294	4:144890294:C:T	rs2323418	T	Intronic	RP11-673E1.1, RP11-673E1.4	0.005
19	45415640	19:45415640:G:A	rs445925	A	Others	APOC1	-0.005
19	11303554	19:11303554:A:G	rs17616661	G	PAVs	KANK2	0.005
1	118154575	1:118154575:T:C	rs67224956	C	Intronic	FAM46C	-0.005
5	173287851	5:173287851:G:A	rs875741	A	Others		-0.005
19	33802542	19:33802542:G:A	rs7251505	A	Others	CTD-2540B15.9	0.005
6	31506691	6:31506691:G:A	rs2071596	A	PAVs	DDX39B	0.005
1	56614762	1:56614762:C:T	rs1230004	T	Intronic	RP1-158P9.1	0.005
8	41507350	8:41507350:C:T	rs7825494	T	Intronic	NKX6-3	-0.004
20	4122485	20:4122485:G:A	rs1764995	A	Intronic	SMOX	0.004
11	100456604	11:100456604:C:T	rs11224302	T	Others		0.004
2	23934087	2:23934087:A:G	rs7563013	G	Others	KLHL29	-0.004
2	8753269	2:8753269:C:T	rs12993630	T	Intronic	AC011747.6	-0.004
6	110760008	6:110760008:A:G	rs12210538	G	PAVs	SLC22A16	0.004
6	25918225	6:25918225:T:C	rs80215559	C	Intronic	SLC17A2	0.004
2	218674697	2:218674697:C:T	rs918949	T	PAVs	TNS1	0.004
X	40861569	X:40861569:C:T	rs5918084	T	Others		-0.004
2	27730940	2:27730940:T:C	rs1260326	C	PAVs	GCKR	-0.004
10	25207403	10:25207403:A:C	rs10828724	C	Intronic	PRTFDC1	-0.004
1	118142444	1:118142444:C:T	rs10489819	T	Intronic	AL157902.3	-0.004
11	9769562	11:9769562:C:G	rs415895	G	PAVs	SWAP70	0.004
1	22038094	1:22038094:G:A	rs1874794	A	Intronic	USP48	0.004
6	135381351	6:135381351:A:G	rs11759077	G	PTVs	CTA-212D2.2	0.004
8	9030160	8:9030160:G:A	rs3748136	A	Others	RP11-10A14.4	0.004
17	31184631	17:31184631:T:C	rs17183628	C	Intronic	MYO1D	0.004
3	24350811	3:24350811:A:G	rs9310736	G	Intronic	THRB	0.004
17	40063083	17:40063083:A:C	rs11079024	C	Intronic	ACLY	0.004
21	30339120	21:30339120:C:A	rs34191159	A	PAVs	LTN1	-0.004
19	46181392	19:46181392:G:C	rs1800437	C	PAVs	GIPR	-0.004
8	116670347	8:116670347:C:T	rs3808477	T	Intronic	TRPS1	-0.004
12	120863422	12:120863422:G:A	rs4767891	A	Others		-0.004
17	76119293	17:76119293:C:T	rs3794738	T	Intronic	TMC6	0.004
16	240895	16:240895:A:G	rs1203956	G	Intronic	LUC7L	0.004
12	95830338	12:95830338:C:T	rs159853	T	PTVs	RP11-167N24.3	-0.004
19	1163934	19:1163934:C:T	rs10853952	T	Intronic	SBNO2	-0.004
17	33872407	17:33872407:G:A	rs11080354	A	Others	RP11-1094M14.10	0.004
9	33838475	9:33838475:T:C	rs2248910	C	Intronic	UBE2R2	0.004
10	71093392	10:71093392:C:T	rs16926246	T	Intronic	HK1	0.004
14	35259151	14:35259151:G:A	rs4982211	A	Intronic	BAZ1A	-0.003
17	44067400	17:44067400:T:C	rs10445337	C	PAVs	MAPT	0.003
2	8750266	2:8750266:A:G	rs3856447	G	Intronic	AC011747.6	-0.003
1	150940625	1:150940625:T:G	rs267738	G	PAVs	CERS2	0.003
17	33946107	17:33946107:A:G	rs225245	G	Intronic	AP2B1	-0.003
22	37469590	22:37469590:C:T	rs387907018	T	PAVs	TMPRSS6	0.003
14	103113552	14:103113552:G:A	rs714827	A	Intronic	RCOR1	-0.003
10	71100726	10:71100726:A:G	rs16926249	G	Intronic	HK1	0.003
20	55975714	20:55975714:T:C	rs6014987	C	Intronic	RBM38	0.003
17	76121864	17:76121864:A:G	rs2748427	G	PAVs	TMC6	-0.003
19	869198	19:869198:C:T	rs55639032	T	Intronic	MED16	-0.003
4	144621779	4:144621779:A:G	rs55935372	G	PAVs	FREM3	-0.003
19	55536595	19:55536595:G:A	rs1613662	A	PAVs	GP6	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 11641 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.