Abstract
Polygenic risk prediction is a widely investigated topic because of its promising clinical applications. Genetic variants in functional regions of the genome are enriched for complex trait heritability. Here, we introduce a method for polygenic prediction, LDpredfunct, that leverages traitspecific functional priors to increase prediction accuracy. We fit priors using the recently developed baselineLD model, including coding, conserved, regulatory, and LDrelated annotations. We analytically estimate posterior mean causal effect sizes and then use crossvalidation to regularize these estimates, improving prediction accuracy for sparse architectures. We applied LDpredfunct to predict 21 highly heritable traits in the UK Biobank (avg N = 373 K as training data). LDpredfunct attained a +4.6% relative improvement in average prediction accuracy (avg prediction R^{2} = 0.144; highest R^{2} = 0.413 for height) compared to SBayesR (the best method that does not incorporate functional information). For height, metaanalyzing training data from UK Biobank and 23andMe cohorts (N = 1107 K) increased prediction R^{2} to 0.431. Our results show that incorporating functional priors improves polygenic prediction accuracy, consistent with the functional architecture of complex traits.
Similar content being viewed by others
Introduction
Genetic variants in functional regions of the genome are enriched for complex trait heritability^{1,2,3,4,5,6}. In this study, we aim to leverage functional priors to improve polygenic prediction^{7,8}. Several studies have shown that incorporating prior distributions on causal effect sizes can improve prediction accuracy^{9,10,11,12,13,14,15,16}, compared to standard Best Linear Unbiased Prediction (BLUP) or Pruning + Thresholding methods^{17,18,19,20,21,22}. Recent efforts to incorporate functional information have produced promising results^{23,24} (see P + TfunctLASSO and AnnoPred results in all main figures below), but maybe limited by dichotomizing between functional and nonfunctional variants^{23} or restricting their analyses to genotyped variants^{24}.
Here, we introduce a method, LDpredfunct, for leveraging traitspecific functional priors to increase polygenic prediction accuracy. We fit functional priors using our recently developed baselineLD model^{25}, which includes coding, conserved, regulatory, and LDrelated annotations. LDpredfunct first analytically estimates posterior mean causal effect sizes, accounting for functional priors and LD between variants. LDpredfunct then uses crossvalidation within validation samples to regularize causal effect size estimates in bins of different magnitude, improving prediction accuracy for sparse architectures. We show that LDpredfunct attains higher polygenic prediction accuracy than other methods in simulations with real genotypes, analyses of 21 highly heritable UK Biobank traits, and metaanalyses of height using training data from UK Biobank and 23andMe cohorts.
Results
Simulations
We performed simulations using real genotypes from the UK Biobank interim release and simulated phenotypes (see Methods). We simulated quantitative phenotypes with SNPheritability \({h}_{g}^{2}=0.5\), using 476,613 imputed SNPs from chromosome 1. We selected either 2000 or 5000 variants to be causal; we refer to these as sparse and polygenic architectures, respectively. We sampled normalized causal effect sizes from normal distributions with variances based on expected causal perSNP heritabilities under the baselineLD model^{25}, fit using stratified LD score regression (SLDSC)^{5,25} applied to height summary statistics from Britishancestry samples from the UK Biobank interim release. We randomly selected 10,000, 20,000, or 50,000 unrelated Britishancestry samples as training samples, and we used 7585 unrelated samples of nonBritish European ancestry as validation samples. By restricting simulations to chromosome 1 (≈1/10 of SNPs), we can extrapolate results to larger sample sizes (≈10× larger; see Application to 21 UK Biobank traits), analogous to the previous work^{16}.
We compared prediction accuracies (R^{2}) for seven main methods: P + T^{18,19}, LDpred^{16}, SBayesR^{9}, P + TfunctLASSO^{23}, AnnoPred^{24}, LDpredfunctinf and LDpredfunct (see Methods). Results are reported in Fig. 1 (main simulations) and Supplementary Fig. 1 (additional values of the number of causal variants); numerical results are reported in Supplementary Tables 1 and 2. Among methods that do not use functional information, the prediction accuracy of LDpred was higher than P + T (particularly for the polygenic architecture), consistent with previous work^{8,16} (see Supplementary Tables 3 and 4 for optimal tuning parameters; surprisingly, at = 50 K training samples, LDpred is optimized by assuming that 100% of SNPs are causal). SBayesR attained a substantial improvement vs. LDpred at N = 10 K training samples (+19% relative improvement for sparse architecture and +8.6% relative improvement for polygenic architecture) but attained prediction R^{2} close to 0 at larger sample sizes (N = 20 K and N = 50 K), perhaps because the algorithm failed to converge (Supplementary Table 1; results not included in Fig. 1).
Incorporating functional information via LDpredfunctinf (a method that does not model sparsity) produced improvements that varied with sample size (+4.7% relative improvement for sparse architecture and +4.8% relative improvement for polygenic architecture at N = 50 K training samples, compared to LDpred; smaller improvements at smaller sample sizes). These results are consistent with the fact that LDpred is known to be sensitive to model assumptions at large sample sizes^{16}. Accounting for sparsity using LDpredfunct further improved prediction accuracy, particularly for the sparse architecture (+7.3% relative improvement for sparse architecture and +5.4% relative improvement for polygenic architecture at N = 50 K training samples, compared to LDpred; smaller improvements at smaller sample sizes). LDpredfunct attained substantially higher prediction accuracy than P + TfunctLASSO in most settings (+11% relative improvement for sparse architecture and +18% relative improvement for polygenic architecture at N = 50 K training samples; smaller improvements at smaller sample sizes). LDpredfunct also attained higher prediction accuracy than AnnoPred at large sample sizes (+5.7% relative improvement for sparse architecture and +3.7% relative improvement for polygenic architecture at N = 50 K training samples; smaller differences at smaller sample sizes) (see Supplementary Table 5 for optimal tuning parameters; surprisingly, at N = 50 K training samples, AnnoPred is optimized by assuming that 100% of SNPs are causal, analogous to LDpred). The difference in prediction accuracy between LDpred and each other method, as well as the difference in prediction accuracy between LDpredfunct and each other method, was statistically significant in most cases (see Supplementary Table 2 e.g. vs. AnnoPred: P < 10^{−125} for sparse architecture and P < 10^{−75} for polygenic architecture at N = 50 K training samples). Simulations with 1000 or 10,000 causal variants generally recapitulated these findings, although SBayesR, P + TfunctLASSO and AnnoPred performed better than LDpredfunct for the very sparse architecture at N = 10 K (Supplementary Table 1).
The average running time for all 7 methods is reported in Supplementary Table 6. We separately report the time to estimate posterior mean causal effect sizes, and the time to compute LD matrices (not applicable for LDpredfunctinf and LDpredfunct) (we do not include the time to compute polygenic risk scores, which is small in comparison and depends on the number of validation samples). For the two methods with the highest prediction R^{2} in analyses of real UK Biobank traits (LDpredfunct and AnnoPred; see below), the average running time was 71 min for LDpredfunct vs. 5249 min for AnnoPred, not including the time to compute LD matrices.
We performed four secondary analyses. First, we assessed the calibration of each method by checking whether regression of true vs. predicted phenotype yielded a slope of 1. We determined that LDpredfunct was wellcalibrated (regression slope 0.98–0.99), LDpred and AnnoPred were fairly wellcalibrated (regression slope 0.85–1.00), and other methods were not wellcalibrated (Supplementary Table 7). Second, we assessed the sensitivity of LDpredfunct to the choice of K = 40 posterior mean causal effect size bins to regularize effect sizes in our main simulations. We determined that results were not sensitive to this parameter (Supplementary Table 8); slightly higher values of K performed slightly better, but we did not finely optimize this parameter. Third, we evaluated a cheating version of LDpredfunct that utilized the true baselineLD model parameters used to simulate the data, instead of estimating these parameters from the data (LDpredfunctcheat). LDpredfunctcheat performed only slightly better than LDpredfunct, indicating that LDpredfunct is not sensitive to the imperfect estimation of functional enrichment parameters (see Supplementary Table 9). Fourth, we simulated traits with lower SNPheritability (\({h}_{g}^{2}=0.25\)) (see Supplementary Table 10). We determined that the improvements attained by LDpredfunct were smaller in these simulations (e.g. +6.9% relative improvement vs. AnnoPred and −1.0% relative improvement vs. LDpred for sparse architecture, +3.4% improvement vs. AnnoPred and +0.6% relative improvement vs. LDpred for polygenic architecture at N = 50 K training samples; smaller improvements at smaller sample sizes).
Application to 21 UK Biobank traits
We applied P + T, LDpred, SBayesR, P + TfunctLASSO, AnnoPred, LDpredfunctinf, and LDpredfunct to 21 UK Biobank traits (14 quantitative traits and 7 binary traits; Supplementary Tables 11 and 12). We analyzed training samples of Britishancestry (avg N = 373 K) and validation samples of nonBritish European ancestry (avg N = 22 K). We included 6,334,603 imputed SNPs in our analyses (see Methods). We computed summary statistics and \({h}_{g}^{2}\) estimates from training samples using BOLTLMM v2.3^{26} (see Supplementary Table 13). We estimated traitspecific functional enrichment parameters for the baselineLD model^{25} by running SLDSC^{5,25} on these summary statistics. Results for quantitative traits are reported in Fig. 2 and Supplementary Table 14, and results for binary traits are reported in Fig. 3 and Supplementary Table 15. Differences between each main prediction method and either LDpred or LDpredfunct (and blockjackknife standard errors on these differences) are reported in Supplementary Table 16, and averages across all 21 traits for main and secondary prediction methods are reported in Supplementary Table 17.
Among methods that do not use functional information, LDpred outperformed P + T (+18% relative improvement in average prediction R^{2}), consistent with simulations under a polygenic architecture (see Supplementary Tables 18 and 19 for optimal tuning parameters) and with the previous work^{8,16}. LDpred also outperformed LDpredinf, a method that does not model sparsity (see Supplementary Table 17). The exclusion of longrange LD regions (see Methods) was critical to LDpred performance, as running LDpred without excluding longrange LD regions (as implemented in a previous version of this paper^{27}) performed much worse (see Supplementary Table 17). SBayesR outperformed LDpred (+5.3% relative improvement in average prediction R^{2}), with no convergence issues in the full UK Biobank analysis (but see below for 113 K interim UK Biobank analysis); we note that expanding the set of SNPs analyzed worsened the performance of SBayesR (see below).
Incorporating functional information via LDpredfunctinf (a method that does not model sparsity) performed only slightly better than LDpred (+0.9% improvement in average prediction R^{2}), but greatly outperformed LDpredinf (+19% relative improvement, P < 10^{−20} for the difference using twosided ztest based on the blockjackknife standard error in Supplementary Table 20). Accounting for sparsity using LDpredfunct substantially improved prediction accuracy (+10%, +4.6%, +7.4% relative improvements in average prediction R^{2} vs. LDpred, SBayesR, LDpredfunctinf; P < 2 × 10^{−4}, P = 0.04, P < 2 × 10^{−4} for differences using twosided ztest based on the blockjackknife standard error in Supplementary Table 16; average prediction R^{2} = 0.144; highest R^{2} = 0.413 for height), consistent with simulations. The relative improvement in avg prediction R^{2} for LDpredfunct vs. LDpred was +9.7% for quantitative traits (higher prediction R^{2} for 14/14 traits), and +11% for binary traits (higher prediction R^{2} for 5/7 traits). We observed a positive but nonsignificant correlation across traits between \({h}_{g}^{2}\) and relative improvement (Supplementary Fig. 2), perhaps due to the limited number of data points and/or contribution of other factors (e.g. polygenicity). LDpredfunct also performed substantially better than P + TfunctLASSO (+20% relative improvement in avg. prediction R^{2}), consistent with simulations under a polygenic architecture. AnnoPred performed slightly but nonsignificantly worse than LDpredfunct (−2.7% relative change in average prediction R^{2} for AnnoPred vs. LDpredfunct, P = 0.35 for the difference using twosided ztest based on the blockjackknife standard error in Supplementary Table 16; see Supplementary Table 21 for optimal tuning parameters).
In the above experiments, LDpredfunct analyzed 373 K training samples and 22 K validation samples and used 90% of the validation samples to estimate regularization weights (and the remaining validation samples to compute predictions) in each crossvalidation fold. It is possible that incorporating data from an additional 20 K samples could confer an unfair advantage for LDpredfunct compared to other methods. To assess this, we performed three additional experiments. First, we repeated the LDpredfunct analyses using smaller validation sample sizes (as low as 1 K), again using 10fold crossvalidation. We determined that results were little changed (Supplementary Table 22). Second, we repeated the LDpredfunct analyses using only 1 K of the 22 K validation samples to estimate regularization weights and the remaining validation samples to compute predictions. Again, we determined that results were little changed (Supplementary Table 23). As the use of 1 K samples to estimate validation weights is a trivial number of additional samples compared to 373 K training samples. Third, we repeated the analysis using 5 K samples omitted from the set of training samples to estimate regularization weights (we recomputed BOLTLMM association statistics using the reduced set of 404 K training samples) and the full set of 22 K validation samples to compute predictions. Again, we determined that results were little changed (Supplementary Table 24). We conclude from these experiments that LDpredfunct does not owe its advantage to incorporating data from a substantial number of additional samples.
We performed 13 secondary analyses. First, we assessed the calibration of each method by checking whether regression of true vs. predicted phenotype yielded a slope of 1. As in our simulations, we determined that LDpredfunct was wellcalibrated (average regression slope: 0.98), LDpred and AnnoPred were fairly wellcalibrated (average regression slope: 0.89 and 0.83, respectively), and other methods were not wellcalibrated (Supplementary Table 25). Second, we assessed the sensitivity of LDpredfunct to the average value of K = 58 posterior mean causal effect size bins to regularize effect sizes in these analyses (see Eq. (6) and Supplementary Table 13). We determined that results were not sensitive to the number of bins (Supplementary Table 26). Third, we determined that functional enrichment information is far less useful when restricting to genotyped variants (e.g. −6.9% relative change in avg prediction R^{2} for LDpredfunct vs. LDpred when both methods are restricted to typed variants; Supplementary Table 17), likely because tagging variants may not belong to enriched functional annotations. Fourth, we repeated the SBayesR analysis using the 2.9 M SNP set instead of the 1.1 M SNP set (see Methods), but determined that this substantially worsened the performance of SBayesR (Supplementary Table 17). Fifth, we evaluated a modification of P + TfunctLASSO in which different weights were allowed for the two predictors (P + TfunctLASSOweighted; see Methods), but results were little changed (+1.1% relative improvement in avg prediction R^{2} vs. P + TfunctLASSO; Supplementary Table 17). Sixth, we obtained similar results for P + TfunctLASSO when defining the "highprior” (HP) SNP set using the top 5% of SNPs with the highest perSNP heritability, instead of the top 10% (see Supplementary Table 17). Seventh, we determined that incorporating baselineLD model functional enrichments that were metaanalyzed across traits (31 traits from ref. ^{25}), instead of the traitspecific functional enrichments used in our primary analyses, slightly reduced the prediction accuracy of LDpredfunctinf (Supplementary Table 17). Eighth, to assess whether the improvement of LDpredfunct is specific to the 75 functional annotations of the baselineLD model, we implemented an analogous method that uses 75 random annotations (LDpredfunct (random)). We determined that LDpredfunct attained a 13% relative improvement in average prediction R^{2} vs. LDpredfunct (random), which performed similarly to LDpred (3.1% decrease in average prediction R^{2} vs. LDpred) (Supplementary Table 17). This implies that the improvement of LDpredfunct is specific to the 75 functional annotations of the baselineLD model. We further note that a method that does not use functional priors but applies the regularization step of LDpredfunct on top of LDpredinf (LDpredinf + sparsity) performed similarly to LDpredfunct (random) (Supplementary Table 17). Ninth, we determined that using our previous baseline model^{5}, instead of the baselineLD model^{25}, slightly reduced the prediction accuracy of LDpredfunctinf and LDpredfunct (Supplementary Table 17). Tenth, we implemented a method analogous to LDpredfunct that uses functional annotations to restrict to the same set of SNPs with expected perSNP heritability \({\sigma }_{i}^{2} \, > \,0\) (2981,1664,306,498 SNPs depending on the trait; see Methods) but then imposes a constant prior on causal effect sizes (LDpredfunct (constant prior)). We determined that LDpredfunct attained a 4.3% relative improvement in average prediction R^{2} vs. LDpredfunct (constant prior) (Supplementary Table 17), implying that including a prior informed by functional annotations is better than not including a prior informed by functional annotations. In addition, LDpredfunct (constant prior) attained a 5.5% relative improvement in average prediction R^{2} vs. LDpred and a 23% relative improvement in average prediction R^{2} vs. LDpredinf (Supplementary Table 17), confirming that regularizing causal effect size estimates in bins of different magnitude increases prediction accuracy (also see the comparison of LDpredfunct vs. LDpredfunctinf above); in addition, some of the improvement of LDpredfunct derives from the removal of (relatively) uninformative SNPs (10% relative improvement for LDpredfunctinf (constant prior) vs. LDpredinf; Supplementary Table 17). Eleventh, we determined that inferring functional enrichments using only the SNPs that passed QC filters and were used for prediction had no impact on the prediction accuracy of LDpredfunctinf (Supplementary Table 17). Twelveth, we determined that using UK10K (instead of 1000 Genomes) as the LD reference panel had virtually no impact on prediction accuracy (Supplementary Table 17). Thirteenth, we determined that using UK10K (instead of 1000 Genomes) as the LD reference panel had virtually no impact on prediction accuracy (Supplementary Table 17).
Application to height in metaanalysis of UK Biobank and 23andMe cohorts
We applied P + T, LDpredinf, SBayesR, P + TfunctLASSO, AnnoPred, LDpredfunctinf, and LDpredfunct to predict height in a metaanalysis of UK Biobank and 23andMe cohorts (see Methods). Training sample sizes were equal to 408,092 for UK Biobank and 698,430 for 23andMe, for a total of 1,106,522 training samples. For comparison purposes, we also computed predictions using the UK Biobank and 23andMe training data sets individually, as well as a training data set consisting of 113,660 Britishancestry samples from the UK Biobank interim release. (The analysis using the 408,092 UK Biobank training samples was nearly identical to the analysis of Fig. 2, except that we used a different set of 5,957,935 SNPs, for consistency throughout this set of comparisons; see Methods.) We used 24,351 UK Biobank samples of nonBritish European ancestry as validation samples in all analyses.
Results are reported in Fig. 4 and Supplementary Table 27. The relative improvements attained by LDpredfunctinf and LDpredfunct were broadly similar across all four training data sets (also see Fig. 2), implying that these improvements are not specific to the UK Biobank data set. Interestingly, compared to the full UK Biobank training data set (R^{2} = 0.415 for LDpredfunct; slightly different from R^{2} = 0.413 in Fig. 2 due to slightly different SNP set), prediction accuracies were only slightly higher for the metaanalysis training data set (R^{2} = 0.431 for LDpredfunct), and were lower for the 23andMe training data set (R^{2} = 0.344 for LDpredfunct), consistent with the ≈30% higher heritability in UK Biobank as compared to 23andMe and other large cohorts^{25,26,28}; the higher heritability in UK Biobank could potentially be explained by lower environmental heterogeneity. We note that in the metaanalysis, we optimized the metaanalysis weights using validation data (similar to ref. ^{29}), instead of performing a fixedeffect metaanalysis. This approach accounts for differences in heritability as well as sample size, and attained a + 3.3% relative improvement in prediction R^{2} compared to fixedeffects metaanalysis (see Supplementary Table 27). We note that SBayesR performed similarly to LDpred in height analyses with ≥408 K training samples (−10% to +0.2% change in average prediction R^{2}) but attained prediction R^{2} close to 0 in the height analysis with 113 K training samples, perhaps because the algorithm failed to converge (Supplementary Table 27; results not included in Fig. 4).
Discussion
We have shown that leveraging traitspecific functional enrichments inferred by SLDSC with the baselineLD model^{25} substantially improves polygenic prediction accuracy. Across 21 UK Biobank traits, we attained substantial improvements in average prediction R^{2} using a method that leverages functional enrichment and performs an additional regularization step to account for sparsity (LDpredfunct). LDpredfunct attained +10% (P < 2 × 10^{−4}) and +4.6% (P = 0.04) relative improvements compared to LDpred^{16} and SBayesR^{9}, two stateoftheart methods that do not model functional enrichment. Thus incorporating functional annotations improves polygenic prediction accuracy. We note that our main analyses used baselineLD model v1.1, but using the updated baselineLD model v2.1 yields slightly higher prediction R^{2} for LDpredfunctinf and LDpredfunct (Supplementary Table 17).
Two previous studies have highlighted the potential advantages of leveraging functional enrichment to improve prediction accuracy^{23,24}. We included both of these methods in all of our analyses. First, ref. ^{23} introduced a method (which we call P + TfunctLASSO) that corrects marginal effect sizes for winner’s curse using LASSO and incorporates functional data to define highprior and lowprior SNP sets. LDpredfunct attained a +19% average relative improvement vs. P + TfunctLASSO across 21 UK Biobank traits. Second, ref. ^{24} introduced AnnoPred, which uses a Bayesian framework to incorporate functional annotations. AnnoPred models sparsity differently than LDpredfunct, as it uses a pointnormal prior to estimating posterior mean effect sizes via Markov Chain Monte Carlo (MCMC), whereas LDpredfunct performs a regularization step to account for sparsity. We note that ref. ^{24} considered only genotyped variants and binary annotations. As noted above, functional enrichment information is far less useful when restricting to genotyped variants (Supplementary Table 17), likely because tagging variants may not belong to enriched functional annotations; thus, the utility of AnnoPred in more general settings is currently unknown. Here, we determined that AnnoPred performed slightly but nonsignificantly worse than LDpredfunct(−2.3% relative change in average prediction R^{2}; P = 0.35 for difference) across 21 UK Biobank traits, consistent with slightly worse results for AnnoPred in simulations at large sample sizes. We emphasize that our work combines binary and continuousvalued functional annotations to improve polygenic risk prediction using imputed variants.
Our work has several limitations. First, LDpredfunct analyzes summary statistic training data (which are publicly available for a broad set of diseases and traits^{30}), but methods that use raw genotypes/phenotypes as training data have the potential to attain higher accuracy^{26}; incorporating functional enrichment information into prediction methods that use raw genotypes/phenotypes as training data remains a direction for future research. Second, the regularization step employed by LDpredfunct to account for sparsity relies on heuristic crossvalidation instead of inferring posterior mean causal effect sizes under a prior sparse functional model; we made this choice because the appropriate choice of sparse functional model is unclear, and because inference of posterior means via MCMC may be subject to convergence issues. As a consequence, the improvement of LDpredfunct over LDpredfunctinf may be contingent on the number of validation samples available for crossvalidation; in particular, for very small validation samples, the number of crossvalidation bins is equal to 1 (Eq. (6)) and LDpredfunct is identical to LDpredfunctinf. However, we determined that results of LDpredfunct were little changed when restricting to smaller validation sample sizes (as low as 1000; see Supplementary Table 22) or using all 22 K validation samples but using only 1 K samples to estimate validation weights ((Supplementary Table 23); this implies that LDpredfunct does not owe its advantage to incorporating data from a substantial number of additional samples. Third, we have considered only singletrait analyses, but leveraging genetic correlations among traits has considerable potential to improve prediction accuracy^{31,32}. Fourth, we have not considered how to leverage functional enrichment for polygenic prediction in related individuals^{33}. Fifth, we have not thoroughly investigated the application LDpredfunct to polygenic prediction in diverse populations^{29,34,35,36} (for which very similar functional enrichments have been reported^{37,38}), as our simulations focused exclusively on prediction in Europeans. However, we evaluated the performance of LDpredfunct in predicting 21 UK Biobank traits in diverse populations using European training data (as in recent studies^{34,35}). The results were promising, particularly in Africans (+23% vs. LDpred (P < 10^{−5}), +18% vs. SBayesR (P = 0.001); see Supplementary Table 28), for which distinguishing causal vs. noncausal variants is particularly important due to differences in LD vs. Europeans^{39}. A more thorough investigation, e.g. incorporating nonEuropean training data^{29}, is an important direction for future research. Sixth, we have not performed a comprehensive assessment of how much different functional annotation models contribute to improvements in prediction accuracy, which remains an important future direction, particularly as functional annotation models will improve as increasingly rich functional data is generated. Specifically, the improvements in prediction accuracy that we reported are a function of the baselineLD model^{25}, but there are many possible ways to improve this model, e.g. by incorporating tissuespecific enrichments^{1,2,3,4,5,6,40,41,42,43}, modeling MAFdependent architectures^{44,45,46}, and/or employing alternative approaches to modeling LDdependent effects^{47}; we anticipate that future improvements to the baselineLD model will yield even larger improvements in prediction accuracy. As an initial step to explore alternative approaches to modeling LDdependent effects, we repeated our analyses using the baselineLD + LDAK model (introduced in ref. ^{48}), which consists of the baselineLD model plus one additional continuous annotation constructed using LDAK weights^{47}. (Recent work has shown that incorporating LDAK weights increases polygenic prediction accuracy in analyses that do not include the baselineLD model^{49}.) We determined that results were virtually unchanged (avg prediction R^{2} = 0.1350 for baselineLD + LDAK vs. 0.1354 for baselineLD using LDpredfunctinf with UK10K SNPs; see Supplementary Tables 17 and 29). Despite these limitations and open directions for future research, our work demonstrates that leveraging functional enrichment using the baselineLD model substantially improves polygenic prediction accuracy.
Methods
Polygenic prediction methods
We compared 7 main prediction methods: Pruning + Thresholding^{18,19} (P + T), LDpred^{16}, SBayesR^{9}, P + T with functionally informed LASSO shrinkage^{23} (P + TfunctLASSO), AnnoPred^{24}, our LDpredfunctinf method, and our LDpredfunct method; we also included LDpredinf^{16}, which is known to attain lower prediction accuracy than LDpred^{16}, in some of our secondary analyses. P + T, LDpredinf, LDpred, and SBayesR are polygenic prediction methods that do not use functional annotations; we did not include RSS^{12} and SBLUP^{11} methods in our comparisons, because ref. ^{9} reported that SBayesR performed as well or better than both RSS and SBLUP and was more computationally efficient (Fig. 2 and Supplementary Fig. 18 of ref. ^{9}). P + TfunctLASSO is a modification of P + T that corrects marginal effect sizes for winner’s curse, accounting for functional annotations. AnnoPred is which uses a Bayesian framework to incorporate functional annotations. LDpredfunctinf is an improvement of LDpredinf that incorporates functionally informed priors on causal effect sizes. LDpredfunct is an improvement of LDpredfunctinf that uses crossvalidation to regularize posterior mean causal effect size estimates, improving prediction accuracy for sparse architectures. Each method is described in greater detail below. In both simulations and analyses of real traits, we used squared correlation (R^{2}) between predicted phenotype and true phenotype in a heldout set of samples as our primary measure of prediction accuracy.
P + T
The P + T method builds a polygenic risk score (PRS) using a subset of independent SNPs obtained via informed LDpruning^{19} (also known as LDclumping) followed by Pvalue thresholding^{18}. Specifically, the method has two parameters, \({R}_{LD}^{2}\) and P_{T}, and proceeds as follows. First, the method prunes SNPs based on a pairwise threshold \({R}_{LD}^{2}\), removing the less significant SNP in each pair. Second, the method restricts to SNPs with an association Pvalue below the significance threshold P_{T}. Letting M be the number of SNPs remaining after LDclumping, polygenic risk scores (PRS) are computed as
where \({\tilde{\beta }}_{i}\) are normalized marginal effect size estimates and g_{i} is a vector of normalized genotypes for SNP i. The parameters \({R}_{LD}^{2}\) and P_{T} are commonly tuned using validation data to optimize prediction accuracy^{18,19}. While in theory, this procedure is susceptible to overfitting, in practice, validation sample sizes are typically large, and \({R}_{LD}^{2}\) and P_{T} are selected from a small discrete set of parameter choices, so that overfitting is considered to have a negligible effect^{7,18,19,29}. Accordingly, in this work, we consider \({R}_{LD}^{2}\in \{0.1,0.2,0.5,0.8\}\) and P_{T} ∈ {1, 0.3, 0.1, 0.03, 0.01, 0.003, 0.001, 3 ∗ 10^{−4}, 10^{−4}, 3 ∗ 10^{−5}, 10^{−5}, 10^{−6}, 10^{−7}, 10^{−8}}, and we always report results corresponding to the best choices of these parameters. The P + T method is implemented in the PLINK software (see Code availability).
LDpredinf
The LDpredinf method estimates posterior mean causal effect sizes under an infinitesimal model, accounting for LD^{16}. The infinitesimal model assumes that normalized causal effect sizes have prior distribution β_{i} ~ N(0, σ^{2}), where \({\sigma }^{2}={h}_{g}^{2}/M,\,\)\({h}_{g}^{2}\) is the SNPheritability, and M is the number of SNPs. The posterior mean causal effect sizes are
where D is the LD matrix between markers, I is the identity matrix, N is the training sample size, \(\tilde{{{{{{{{\boldsymbol{\beta }}}}}}}}}\) is the vector of marginal association statistics, and \({h}_{l}^{2}\approx k{h}^{2}/M\) is the heritability of the k SNPs in the region of LD; following ref. ^{16} we use the approximation \(1{h}_{l}^{2}\approx 1\), which is appropriate when M > > k. D is typically estimated using validation data, restricting to nonoverlapping LD windows. We used the default LD window size, which is M/3000. \({h}_{g}^{2}\) can be estimated from raw genotype/phenotype data^{26,28} (the approach that we use here; see below), or can be estimated from summary statistics using the aggregate estimator as described in ref. ^{16}. To approximate the normalized marginal effect size ref. ^{16} uses the pvalues to obtain absolute Z scores and then multiplies absolute Z scores by the sign of the estimated effect size. When sample sizes are very large, pvalues may be rounded to zero, in which case we approximate normalized marginal effect sizes \({\widehat{\beta }}_{i}\) by \({\widehat{b}}_{i}\frac{\sqrt{2* {p}_{i}* (1{p}_{i})}}{\sqrt{{\sigma }_{Y}^{2}}}\), where \(\widehat{{b}_{i}}\) is the perallele marginal effect size estimate, p_{i} is the minor allele frequency of SNP i, and \({\sigma }_{Y}^{2}\) is the phenotypic variance in the training data. This applies to all the methods that use normalized effect sizes. Although the published version of LDpred requires a matrix inversion (Eq. (2)), we have implemented a computational speedup that computes the posterior mean causal effect sizes by efficiently solving^{50} the system of linear equations \((\frac{1}{{\sigma }^{2}}{{{{{{{\bf{I}}}}}}}}+N* {{{{{{{\bf{D}}}}}}}})E({{{{{{{\boldsymbol{\beta }}}}}}}} \tilde{{{{{{{{\boldsymbol{\beta }}}}}}}}},{{{{{{{\bf{D}}}}}}}})=N\tilde{{{{{{{{\boldsymbol{\beta }}}}}}}}}\).
LDpred
The LDpred method is an extension of LDpredinf that uses a pointnormal prior to estimating posterior mean effect sizes via Markov Chain Monte Carlo (MCMC)^{16}. It assumes a Gaussian mixture prior: \({\beta }_{i} \sim N(0,{h}_{g}^{2}/M* p)\) with probability p, and β_{i} ~ 0 with probability 1 − p, where p is the proportion of causal SNPs. The method is optimized by considering different values of p (1E−4, 3E−4, 1E−3, 3E−3, 0.01,0.03,0.1,0.3,1); in the special case where 100% of SNPs are assumed to be causal, LDpred is roughly equivalent to LDpredinf. We excluded SNPs from longrange LD regions (reported in ref. ^{51}), as our secondary analyses showed that including these regions were suboptimal, consistent with ref. ^{9}.
SBayesR
The SBayesR method infers posterior mean causal effect sizes from GWAS summary statistics and an LD matrix^{9}. It assumes a finite mixture of normal distributions to account for sparsity, defined as: \({\beta }_{i} \sim N(0,{\gamma }_{c}{h}_{g}^{2})\) with probability π_{c}, where c ranges from 1 to C, the total number of components in the mixture model. We used as input the recommended parameters from ref. ^{9}, with C = 4 mixtures with parameters γ_{c} = (0, 0.01, 0.1, 1.0). The method requires a shrunk LD matrix^{12}. The authors of ref. ^{9} made available shrunk LD matrices estimated from 50,000 randomly selected white British individuals from the UK Biobank^{51} for two different SNPs sets. The 1.1M SNP set consists of 1,094,841 variants, constructed by restricting 1,365,446 SNPs from HapMap3^{52} to MAF > 0.01 and removing strand ambiguous SNPs and longrange LD regions (as reported in ref. ^{51}). The 2.9M SNP set consists of 2,865,810 variants, constructed by applying LDpruning (R^{2} > 0.99) to a larger set of 8 million variants from the UK Biobank^{51} with MAF > 0.01, overlapped with a previous large GWAS^{53} and present in 1000 Genomes^{54}. We note that we could not scale the SBayesR analysis to the full set of 6,334,603 variants used in other analyses due to computational constraints. We used the 1.1M SNP set in our primary analyses as it achieved the highest average prediction R^{2} in our real traits analyses (see Results section), but we also considered the 2.9 M SNP set in secondary analyses. For analyses that use BOLTLMM summary statistics we used N_{effective} as reported in ref. ^{26}.
P + TfunctLASSO
Reference^{23} proposed an extension of P + T that corrects the marginal effect sizes of SNPs for winner’s curse and incorporates external functional annotation data (P + TfunctLASSO). The winner’s curse correction is performed by applying a LASSO shrinkage to the marginal association statistics of the PRS:
where \(\lambda ({P}_{T})={{{\Phi }}}^{1}(1\frac{{P}_{T}}{2})sd({\tilde{\beta }}_{i})\), where Φ^{−1} is the inverse standard normal CDF. Functional annotations are incorporated via two disjoint SNPs sets, representing "highprior” SNPs (HP) and "lowprior” SNPs (LP), respectively. We define the HP SNP set for P + TfunctLASSO as the set of SNPs in the top 10% of expected perSNP heritability under the baselineLD model^{25}, which includes coding, conserved, regulatory, and LDrelated annotations, whose enrichments are jointly estimated using stratified LD score regression^{5,25} (see BaselineLD model annotations section). We also performed secondary analyses using the top 5% (P + TfunctLASSOtop5%). We define PRS_{LASSO,HP}(P_{HP}) to be the PRS restricted to the HP SNP set, and PRS_{LASSO, LP}(P_{LP}) to be the PRS restricted to the LP SNP set, where P_{HP} and P_{LP} are the optimal significance thresholds for the HP and LP SNP sets, respectively. We define PRS_{LASSO}(P_{HP}, P_{LP}) = PRS_{LASSO,HP}(P_{HP}) + PRS_{LASSO,LP}(P_{LP}). We also performed secondary analyses where we allow an additional regularization to the two PRS: PRS_{LASSO}(P_{HP}, P_{LP}) = α_{1}PRS_{LASSO,HP}(P_{HP}) + α_{2}PRS_{LASSO, LP}(P_{LP}); we refer to this method as P + TfunctLASSOweighted.
AnnoPred
AnnoPred^{24} uses a Bayesian framework to incorporate functional priors while accounting for LD, optimizing prediction R^{2} over different assumed values of the proportion of causal SNPs. Reference^{24} proposed two different priors for use with AnnoPred. The first prior assumes the same proportion of causal SNPs but different causal effect size variance across functional annotations, and uses a pointnormal prior to estimating posterior mean effect sizes via Markov Chain Monte Carlo (MCMC). In the special case where 100% of SNPs are assumed to be causal, AnnoPred is roughly equivalent to LDpredfunctinf (see below). The second prior assumes different proportions of causal SNPs but the same causal effect size variance across functional annotations. We only consider the first prior, since the second prior cannot be extended to incorporate continuousvalued annotations from the baselineLD model. We excluded SNPs from longrange LD regions (as reported in ref. ^{51}) when running AnnoPred. We used the default LD window size, which is M/3000.
LDpredfunctinf
We modify LDpredinf to incorporate functionally informed priors on causal effect sizes using the baselineLD model^{25}, which includes coding, conserved, regulatory, and LDrelated annotations, whose enrichments are jointly estimated using stratified LD score regression^{5,25}. Specifically, we assume that normalized causal effect sizes have prior distribution \({\beta }_{i} \sim N(0,c* {\sigma }_{i}^{2})\), where \({\sigma }_{i}^{2}\) is the expected perSNP heritability under the baselineLD model (fit using training data only) and c is a normalizing constant such that \(\mathop{\sum }\nolimits_{i = 1}^{M}{{\mathbb{1}}}_{\{{\sigma }_{i}^{2} \,{ > }\,0\}}c{\sigma }_{i}^{2}={h}_{g}^{2}\); SNPs with \({\sigma }_{i}^{2}\le 0\) are removed, which is equivalent to setting \({\sigma }_{i}^{2}=0\). The posterior mean causal effect sizes are
where M_{+} is the number of SNPs with \({\sigma }_{i}^{2}\, > \, 0\). The posterior mean causal effect sizes are computed by solving the system of linear equations \({{{{{{{\bf{W}}}}}}}}E[{{{{{{{\boldsymbol{\beta }}}}}}}} \tilde{{{{{{{{\boldsymbol{\beta }}}}}}}}},{{{{{{{\bf{D}}}}}}}},{\sigma }_{1}^{2},\ldots ,{\sigma }_{M}^{2}]=N* \tilde{{{{{{{{\boldsymbol{\beta }}}}}}}}}\). \({h}_{g}^{2}\) is estimated as described above (see LDpredinf). D is estimated using validation data, restricting to windows of size 0.15%M_{+}. In principle, it is possible to use banding to define the LD matrices, where LD between distant pairs of SNPs (10 Mb or more) is rounded to zero^{55}, but we elected to use the simpler windowbased approach (as in ref. ^{16}).
LDpredfunct
We modify LDpredfunctinf to regularize posterior mean causal effect sizes using crossvalidation. We rank the SNPs by their (absolute) posterior mean causal effect sizes, partition the SNPs into K bins (analogous to ref. ^{56}) where each bin has roughly the same sum of squared posterior mean effect sizes, and determine the relative weights of each bin based on the predictive value in the validation data. Intuitively if a bin is dominated by noncausal SNPs, the inferred relative weight will be lower than for a bin with a high proportion of causal SNPs. This nonparametric shrinkage approach can optimize prediction accuracy regardless of the genetic architecture. In detail, let \(S={\sum }_{i}E{[{\beta }_{i} {\tilde{\beta }}_{i}]}^{2}\). To define each bin, we first rank the posterior mean effect sizes based on their squared values \(E{[{\beta }_{i} {\tilde{\beta }}_{i}]}^{2}\). We define bin b_{1} as the smallest set of top SNPs with \({\sum }_{i\in {b}_{1}}E{[{\beta }_{i} {\tilde{\beta }}_{i}]}^{2}\ge \frac{S}{K}\), and iteratively define bin b_{k} as the smallest set of additional top SNPs with \({\sum }_{i\in {b}_{1},\ldots ,{b}_{k}}E{[{\beta }_{i} {\tilde{\beta }}_{i}]}^{2}\ge \frac{kS}{K}\). Let \(\,{{\mbox{PRS}}}\,(k)={\sum }_{i\in {b}_{k}}E[{\beta }_{i} {\tilde{\beta }}_{i}]{g}_{i}\). We define
where the binspecific weights α_{k} are optimized using validation data via 10fold crossvalidation. For each heldout fold in turn, we split the data so we estimate the weights α_{k} using the samples from the other nine folds (90% of the validation) and compute PRS on the heldout fold using these weights (10% of the validation); thus, in each crossvalidation fold, the validation samples used to estimate regularization weights are disjoint from the validation samples used to compute predictions. We then compute the average prediction R^{2} across the 10 heldout folds. To avoid overfitting when K is very close to N, we set the number of bins (K) to be between 1 and 100, such that it is proportional to \({h}_{g}^{2}\) and the number of samples used to estimate the K weights in each fold is at least 100 times larger than K:
where N is the number of validation samples. For highly heritable traits (\({h}_{g}^{2} \sim 0.5\)), LDpredfunct reduces to the LDpredfunctinf method if there are ~ 200 validation samples or fewer; for less heritable traits (\({h}_{g}^{2} \sim 0.1\)), LDpredfunct reduces to the LDpredfunctinf method if there are ~ 1000 validation samples or fewer. In simulations, we set K to 40 (based on 7,585 validation samples; see below), approximately concordant with Eq. (6). The value of 100 in the denominator of Eq. (6) was coarsely optimized in simulations, but was not optimized using real trait data. We note that functional annotations are not used in the crossvalidation step (although they do impact the posterior mean causal effect size provided as input to this step). Thus, it is likely that SNPs from a given functional annotation will fall into different bins (possibly all of the bins).
Standard errors. Standard errors for the prediction R^{2} of each method and the difference in prediction R^{2} between two methods were computed via blockjackknife using 200 genomic jackknife blocks^{5}; this is more conservative than computing standard errors based on the number of validation samples, which does not account for variation across a finite number of SNPs. For each method, we first optimized any relevant tuning parameters using the entire genome and then analyzed each jackknife block using those tuning parameters.
Simulations
We simulated quantitative phenotypes using real genotypes from the UK Biobank interim release (see below). We used up to 50,000 unrelated Britishancestry samples as training samples, and 7,585 samples of other European ancestries as validation samples (see below). We made these choices to minimize confounding due to shared population stratification or cryptic relatedness between training and validation samples (which, if present, could overstate the prediction accuracy that could be obtained in independent samples^{57}), while preserving a large number of training samples. We restricted our simulations to 459,284 imputed SNPs on chromosome 1 (see below), fixed the number of causal SNPs at 2,000 or 5,000 (we also performed secondary simulations with 1000 or 10,000 causal variants), and fixed the SNPheritability \({h}_{g}^{2}\) at 0.5. We sampled normalized causal effect sizes β_{i} for causal SNPs from a normal distribution with variance equal to \(\frac{{\sigma }_{i}^{2}}{p}\), where p is the proportion of causal SNPs and \({\sigma }_{i}^{2}\) is the expected causal perSNP heritability under the baselineLD model^{25}, fit using stratified LD score regression (SLDSC)^{5,25} applied to height summary statistics computed from unrelated Britishancestry samples from the UK Biobank interim release (N = 113,660). We computed perallele effect sizes b_{i} as \({b}_{i}=\frac{{\beta }_{i}}{\sqrt{2{p}_{i}(1{p}_{i})}}\), where p_{i} is the minor allele frequency for SNP i estimated using the validation genotypes. We simulated phenotypes as \({Y}_{j}=\mathop{\sum }\nolimits_{i}^{M}{b}_{i}{g}_{ij}+{\epsilon }_{j}\), where \({\epsilon }_{j} \sim N(0,1{h}_{g}^{2})\). We set the training sample size to either 10,000, 20,000, or 50,000. The motivation to perform simulations using one chromosome is to be able to extrapolate performance at larger sample sizes^{16} according to the ratio N/M, where N is the training sample size. We compared each of the seven methods described above. For LDpredfunctinf and LDpredfunct, for each simulated trait we used SLDSC (applied to training data only) to estimate baselineLD model parameters. For LDpredfunct, we report R^{2} as the average prediction R^{2} across the 10 heldout folds.
Full UK Biobank data set
The full UK Biobank data set includes 459,327 Europeanancestry samples and ~ 20 million imputed SNPs^{51} (after filtering as in ref. ^{26}, excluding indels and structural variants). We selected 21 UK Biobank traits (14 quantitative traits and 7 binary traits) with phenotyping rate >80% (>80% of females for age at menarche, >80% of males for balding), SNPheritability \({h}_{g}^{2} \, > \, 0.2\) for quantitative traits, observedscale SNPheritability \({h}_{g}^{2} \, > \, 0.1\) for binary traits, and low correlation between traits (as described in ref. ^{26}). We restricted training samples to 409,728 Britishancestry samples^{51}, including related individuals (avg N = 373 K phenotyped training samples; see Supplementary Table 11 for quantitative traits and Supplementary Table 12 for binary traits). We computed association statistics from training samples using BOLTLMM v2.3^{26}. We have made these association statistics publicly available (see Data availability). We restricted validation samples to 24,436 samples of nonBritish European ancestry, after removing validation samples that were related ( >0.05) to training samples and/or other validation samples (avg N = 22 K phenotyped validation samples; see Supplementary Tables 11 and 12). As in our simulations, we made these choices to minimize confounding due to shared population stratification or cryptic relatedness between training and validation samples (which, if present, could overstate the prediction accuracy that could be obtained in independent samples^{57}), while preserving a large number of training samples. We analyzed 6,334,603 genomewide imputed SNPs, after removing SNPs with minor allele frequency <1%, removing SNPs with imputation accuracy <0.9, and removing A/T and C/G SNPs to eliminate potential strand ambiguity. We used \({h}_{g}^{2}\) estimates from BOLTLMM v2.3^{26} as input to LDpred, AnnoPred, LDpredfunctinf, and LDpredfunct.
UK Biobank interim release
The UK Biobank interim release includes 145,416 Europeanancestry samples^{58}. We used the UK Biobank interim release both in simulations using real genotypes, and in a subset of analyses of height phenotypes (to investigate how prediction accuracy varies with training sample size).
In our analyses of height phenotypes, we restricted training samples to 113,660 unrelated (≤0.05) Britishancestry samples for which height phenotypes were available. We computed association statistics by adjusting for 10 PCs^{59}, estimated using FastPCA^{60} (see Code availability). For consistency, we used the same set of 24,351 validation samples of nonBritish European ancestry with height phenotypes as defined above. We analyzed 5,957,957 genomewide SNPs, after removing SNPs with minor allele frequency <1%, removing SNPs with imputation accuracy <0.9, removing SNPs that were not present in the 23andMe height data set (see below), and removing A/T and C/G SNPs to eliminate potential strand ambiguity.
In our simulations, we restricted training samples to up to 50,000 of the 113,660 unrelated Britishancestry samples, and restricted validation samples to 8441 samples of nonBritish European ancestry, after removing validation samples that were related ( >0.05) to training samples and/or other validation samples. We restricted the 5,957,957 genomewide SNPs (see above) to chromosome 1, yielding 459,284 SNPs after QC.
23andMe height summary statistics
The 23andMe data set consists of summary statistics computed from 698,430 Europeanancestry samples (23andMe customers who consented to participate in research) at 9,898,287 imputed SNPs, after removing SNPs with minor allele frequency <1% and that passed QC filters (which include filters on imputation quality, avg.rsq <0.5 or min.rsq <0.3 in any imputation batch, and imputation batch effects). Analyses were restricted to the set of individuals with >97% European ancestry, as determined via an analysis of local ancestry^{61}. Summary association statistics were computed using linear regression adjusting for age, gender, genotyping platform, and the top five principal components to account for residual population structure. The summary association statistics will be made available to qualified researchers (see Data availability).
We analyzed 5,808,258 genomewide SNPs, after removing SNPs with minor allele frequency <1%, removing SNPs with imputation accuracy <0.9, removing SNPs that were not present in the full UK Biobank data set (see above), and removing A/T and C/G SNPs to eliminate potential strand ambiguity.
Metaanalysis of full UK Biobank and 23andMe height data sets
We metaanalyzed height summary statistics from the full UK Biobank and 23andMe data sets. We define
where PRS_{i} is the PRS obtained using training data from cohort i. The PRS can be obtained using P + T, P + TfunctLASSO, LDpredinf, or LDpredfunct. The metaanalysis weights γ_{i} can either be specified via fixedeffect metaanalysis (e.g. \({\gamma }_{i}=\frac{{N}_{i}}{\sum {N}_{i}}\)) or optimized using validation data^{29}. We use the latter approach, which can improve prediction accuracy (e.g. if the cohorts differ in their heritability as well as their sample size). In our primary analyses, we fit the weights γ_{i} insample and report prediction accuracy using adjusted R^{2} to account for insample fitting^{29}. We also report results using 10fold crossvalidation: for each heldout fold in turn, we estimate the weights γ_{i} using the other nine folds and compute PRS on the heldout fold using these weights. We then compute the average prediction R^{2} across the 10 heldout folds.
When using LDpredfunct as the prediction method, we perform the metaanalysis as follows. First, we use LDpredfunctinf to fit metaanalysis weights γ_{i}. Then, we use γ_{i} to compute (metaanalysis) weighted posterior mean causal effect sizes (PMCES) via PMCES = γ_{1}PMCES_{1} + γ_{2}PMCES_{2}, which are binned into k bins. Then, we estimate binspecific weights α_{k} (used to compute (metaanalysis + binspecific) weighted posterior mean causal effect sizes \(\mathop{\sum }\nolimits_{k = 1}^{K}{\alpha }_{k}\,{{\mbox{PMCES}}}\,(k)\)) using validation data via 10fold crossvalidation.
BaselineLD model annotations
The baselineLD model (v1.1) contains a broad set of 75 functional annotations (including coding, conserved, regulatory, and LDrelated annotations), whose enrichments are jointly estimated using stratified LD score regression^{5,25}. For each trait, we used the τ_{c} values estimated for that trait to compute \({\sigma }_{i}^{2}\), the expected perSNP heritability of SNP i under the baselineLD model, as
where a_{c}(i) is the value of annotation c at SNP i.
Joint effect sizes τ_{c} for each annotation c are estimated via
where l(i, c) is the LD score of SNP i with respect to annotation a_{c} and \({\chi }_{i}^{2}\) is the chisquare statistic for SNP i. We note that τ_{c} quantifies effects that are unique to annotation c. In all analyses of real phenotypes, τ_{c} and \({\sigma }_{i}^{2}\) were estimated using training samples only.
In our primary analyses, we used 489 unrelated European samples from phase 3 of the 1000 Genomes Project^{54} as the reference data set to compute LD scores, as in ref. ^{25}.
To verify that our 1000 Genomes reference data set produces reliable LD estimates, we repeated our LDpredfunct analyses using SLDSC with 3,567 unrelated individuals from UK10K^{62} as the reference data set (as in ref. ^{48}), ensuring a closer ancestry match with Britishancestry UK Biobank samples. We also repeated our LDpredfunct analyses using SLDSC with the baselineLD + LDAK model (instead of the baselineLD model), with UK10K as the reference data set. The baselineLD + LDAK model (introduced in ref. ^{48}) consists of the baselineLD model plus one additional continuous annotation constructed using LDAK weights^{47}, which has values \({({p}_{j}(1{p}_{j}))}^{1+\alpha }{w}_{j}\), where α = −0.25, p_{j} is the allele frequency of SNP j, and w_{j} is the LDAK weight of SNP j computed using UK10K data.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Data availability
Source data are provided with this paper. We used BOLTLMM v2.3 association statistics: https://data.broadinstitute.org/alkesgroup/UKBB/UKBB_409K/. The baselineLD annotations (v.2.1) used to compute functional enrichments in the primary analysis are available at https://alkesgroup.broadinstitute.org/LDSCORE/1000G_Phase3_baseline_v1.2_ldscores.tgz. 1000 Genomes Project, http://www.1000genomes.org/. Access to the UK10K data used in the secondary analysis is available via application in https://www.uk10k.org/data_access.html. Access to the UK Biobank resource is available via application in http://www.ukbiobank.ac.uk/. 23andMe height association statistics: The full summary statistics for the 23andMe height GWAS data will have restricted access, and will be made available through 23andMe to qualified researchers under an agreement with 23andMe that protects the privacy of the 23andMe participants. Please visit https://research.23andme.com/collaborate/#publicationfor more information and to apply to access the data. SBayesR shrunk and sparse LD matrices can be downloaded from Zenodo public repositoryhttps://zenodo.org/, for both 1.09 million HapMap3 (https://doi.org/10.5281/zenodo.3350914) and 2.8 million pruned variants (https://doi.org/10.5281/zenodo.3375373). Source data are provided with this paper.
Code availability
Software implementing the LDpredfunctinf and LDpredfunct^{63}: https://www.hsph.harvard.edu/alkesprice/software (https://doi.org/10.5281/zenodo.4579879). LDscore regression v1.0.1 software: https://github.com/bulik/ldsc. BOLTLMM v2.3 software http://data.broadinstitute.org/alkesgroup/BOLTLMM/. FASTPCA is available in EIGENSOFT(7.2.1) at https://github.com/DReichLab/EIG/archive/v7.2.1.tar.gz (more details in https://www.hsph.harvard.edu/alkesprice/software). AnnoPred: https://github.com/yiminghu/AnnoPred. SBayesR 2.0 software: http://cnsgenomics.com/software/gctb/. LDAK version 5 is available at http://dougspeed.com/downloads/. Plink 2.0 is available at: https://www.coggenomics.org/plink/2.0/.
References
Maurano, M. T. et al. Systematic localization of common diseaseassociated variation in regulatory DNA. Science 337, 1222794 (2012).
Trynka, G. et al. Chromatin marks identify critical cell types for fine mapping complex trait variants. Nat. Genet. 45, 124 (2013).
Pickrell, J. K. Joint analysis of functional genomic data and genomewide association studies of 18 human traits. Am. J. Hum Genet. 94, 559–573 (2014).
Roadmap Epigenomics Consortium. et al. Integrative analysis of 111 reference human epigenomes. Nature 518, 317–330 (2015).
Finucane, H. K. et al. Partitioning heritability by functional annotation using genomewide association summary statistics. Nat. Genet. 47, 1228–1235 (2015).
Farh, K. K.H. et al. Genetic and epigenetic fine mapping of causal autoimmune disease variants. Nature 518, 337 (2015).
Chatterjee, N., Shi, J. & GarcíaClosas, M. Developing and evaluating polygenic risk prediction models for stratified disease prevention. Nat. Rev. Genet. 17, 392–406 (2016).
Khera, A. V. et al. Genomewide polygenic scores for common diseases identify individuals with risk equivalent to monogenic mutations. Nat. Genet. 50, 1219–1224 (2018).
LloydJones, L. R. et al. Improved polygenic prediction by bayesian multiple regression on summary statistics. Nat. Commun. 10, 5086 (2019).
Ge, T., Chen, C.Y., Ni, Y., Feng, Y.C. A. & Smoller, J. W. Polygenic prediction via Bayesian regression and continuous shrinkage priors. Nat. Commun. 10, 1776 (2019).
Maier, R. M. et al. Improving genetic prediction by leveraging genetic correlations among human diseases and traits. Nat. Commun. 9, 989 (2018).
Zhu, X. & Stephens, M. Bayesian largescale multiple regression with summary statistics from genomewide association studies. The Ann Appl Stat. 11, 1561 (2017).
Zhou, X., Carbonetto, P. & Stephens, M. Polygenic modeling with bayesian sparse linear mixed models. PLOS Genet. 9, 1–14 (2013).
Moser, G. et al. Simultaneous discovery, estimation and prediction analysis of complex traits using a bayesian mixture model. PLOS Genet. 11, 1–22 (2015).
Speed, D. & Balding, D. J. Multiblup: improved SNPbased prediction for complex traits. Genome Res. 24, 1550–1557 (2014).
Vilhjálmsson, B. J. et al. Modeling linkage disequilibrium increases accuracy of polygenic risk scores. Am J. Hum Genet. 97, 576–592 (2015).
Henderson, C. R. Best linear unbiased estimation and prediction under a selection model. Biometrics 31, 423–447 (1975).
International Schizophrenia Consortium. et al. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460, 748–752 (2009).
Stahl, E. A. et al. Bayesian inference analyses of the polygenic architecture of rheumatoid arthritis. Nat. Genet 44, 483–489 (2012).
Choi, S. W. & O’Reilly, P. F. Prsice2: Polygenic risk score software for biobankscale data. Gigascience 8, giz082 (2019).
Mak, T. S. H., Porsch, R. M., Choi, S. W., Zhou, X. & Sham, P. C. Polygenic scores via penalized regression on summary statistics. Genet. Epidemiol 41, 469–480 (2017).
Newcombe, P. J., Nelson, C. P., Samani, N. J. & Dudbridge, F. A flexible and parallelizable approach to genomewide polygenic risk scores. Genet. Epidemiol. 43, 730–741 (2019).
Shi, J. et al. Winner’s curse correction and variable thresholding improve performance of polygenic risk modeling based on genomewide association study summarylevel data. PLOS Genet. 12, e1006493 (2016).
Hu, Y. et al. Leveraging functional annotations in genetic risk prediction for human complex diseases. PLOS Comput. Biol.13, 1–16 (2017).
Gazal, S. et al. Linkage disequilibriumdependent architecture of human complex traits shows action of negative selection. Nature Genetics 49, 1421 EP– (2017).
Loh, P.R., Kichaev, G., Gazal, S., Schoech, A. P. & Price, A. L. Mixedmodel association for biobankscale datasets. Nat. Genet. 50, 906–908 (2018).
MárquezLuna, C. et al. Modeling functional enrichment improves polygenic prediction accuracy in UK biobank and 23andme data sets. Preprint at bioRxiv https://doi.org/10.1101/375337 (2018).
Ge, T., Chen, C.Y., Neale, B. M., Sabuncu, M. R. & Smoller, J. W. Phenomewide heritability analysis of the UK Biobank. PLOS Genetics 13, e1006711 (2017).
MárquezLuna, C., Loh, P.R., South Asian Type 2 Diabetes (SAT2D) Consortium, The SIGMA Type 2 Diabetes Consortium & Price, A. L. Multiethnic polygenic risk scores improve risk prediction in diverse populations. Genet. Epidemiol. 41, 811–823 (2017).
Pasaniuc, B. & Price, A. L. Dissecting the genetics of complex traits using summary association statistics. Nat. Rev. Genet. 18, 117 (2017).
Maier, R. et al. Joint analysis of psychiatric disorders increases accuracy of risk prediction for schizophrenia, bipolar disorder, and major depressive disorder. Am. J. Hum. Genet. 96, 283–294 (2015).
Maier, R. M. et al. Improving genetic prediction by leveraging genetic correlations among human diseases and traits. Nat. Commun. 9, 989 (2018).
Tucker, G. et al. Twovariancecomponent model improves genetic prediction in family datasets. Am. J. Hum. Genet. 97, 677–690 (2015).
Martin, A. R. et al. Clinical use of current polygenic risk scores may exacerbate health disparities. Nat. Genet. 51, 584–591 (2019).
Duncan, L. et al. Analysis of polygenic risk score usage and performance in diverse human populations. Nat. Commun. 10, 3328 (2019).
Gurdasani, D., Barroso, I., Zeggini, E. & Sandhu, M. S. Genomics of disease risk in globally diverse populations. Nat. Rev Genet. 20, 520–535 (2019).
Kichaev, G. & Pasaniuc, B. Leveraging functionalannotation data in transethnic finemapping studies. Am. J. Hum Genet. 97, 260–271 (2015).
Kanai, M. et al. Genetic analysis of quantitative traits in the Japanese population links cell types to complex human diseases. Nat. Genet. 50, 390–400 (2018).
Wang, Y. et al. Theoretical and empirical quantification of the accuracy of polygenic scores in ancestry divergent populations. Nat. Commun. 11, 3865 (2020).
Calderon, D. et al. Inferring relevant cell types for complex traits by using singlecell gene expression. Am. J. Hum. Genet. 101, 686–699 (2017).
Ongen, H. et al. Estimating the causal tissues for complex traits and diseases. Nat. Genet. 49, 1676–1683 (2017).
Finucane, H. K. et al. Heritability enrichment of specifically expressed genes identifies diseaserelevant tissues and cell types. Nat. Genet. 50, 621–629 (2018).
Backenroth, D. et al. FUNLDA: A Latent dirichlet allocation model for predicting tissuespecific functional effects of noncoding variation: methods and applications. Am. J. Hum. Genet. 102, 920–942 (2018).
Zeng, J. et al. Signatures of negative selection in the genetic architecture of human complex traits. Nat. Genet. 50, 746–753 (2018).
Gazal, S. et al. Functional architecture of lowfrequency variants highlights strength of negative selection across coding and noncoding annotations. Nat. Genet. 50, 1600–1607 (2018).
Schoech, A. P. et al. Quantification of frequencydependent genetic architectures in 25 uk biobank traits reveals action of negative selection. Nat. Commun. 10, 790 (2019).
Speed, D. et al. Reevaluation of snp heritability in complex human traits. Nat. Genet. 49, 986 (2017).
Gazal, S., MarquezLuna, C., Finucane, H. K. & Price, A. L. Reconciling sldsc and ldak functional enrichment estimates. Nat. Genet. 51, 1202–1204 (2019).
Speed, D. & Balding, D. J. Sumher better estimates the snp heritability of complex traits from summary statistics. Nat. Genet. 51, 277–284 (2019).
Strang, G. Linear Algebra and Its Applications, 2nd edn (Academic Press, Inc., 1980).
Bycroft, C. et al. The UK biobank resource with deep phenotyping and genomic data. Nature 562, 203–209 (2018).
International HapMap 3 Consortium. et al. Integrating common and rare genetic variation in diverse human populations. Nature 467, 52 (2010).
Lee, J. J. et al. Gene discovery and polygenic prediction from a genomewide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018).
1000 Genomes Project Consortium. et al. A global reference for human genetic variation. Nature 526, 68–74 (2015).
Yang, J. et al. Conditional and joint multiplesnp analysis of gwas summary statistics identifies additional variants influencing complex traits. Nat. Genet. 44, 369–375 (2012).
Chun, S. et al. Nonparametric polygenic risk prediction using partitioned GWAS summary statistics. Am J. Hum Genet. 107, 4659 (2019).
Wray, N. R. et al. Pitfalls of predicting complex traits from snps. Nat. Rev. Genet. 14, 507 EP– (2013).
Sudlow, C. et al. Uk biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 12, e1001779 (2015).
Galinsky, K. J., Loh, P.R., Mallick, S., Patterson, N. J. & Price, A. L. Population structure of uk biobank and ancient eurasians reveals adaptation at genes influencing blood pressure. Am. J. Hum Genet. 99, 1130–1139 (2016).
Galinsky, K. J. et al. Fast principalcomponent analysis reveals convergent evolution of ADH1b in Europe and East Asia. The Am. J. Hum Genet. 98, 456–472 (2016).
Durand, E. Y., Do, C. B., Mountain, J. L. & Macpherson, J. M. Ancestry composition: A novel, efficient pipeline for ancestry deconvolution. Preprint at bioRxiv https://doi.org/10.1101/010512 (2014).
UK10K Consortium. et al. The uk10k project identifies rare variants in health and disease. Nature 526, 82 (2015).
MarquezLuna, C. Incorporating functional priors improves polygenic prediction accuracy in UK Biobank and 23andMe data sets. carlaml/LDpredfunct: First release of LDpredfunct software, https://doi.org/10.5281/zenodo.4579879 (2021).
Acknowledgements
We thank the research participants and employees of 23andMe for making this work possible. We are grateful to S. Sunyaev, S. Chun, L. O’Connor, O. Weissbrod, and H. Finucane for helpful discussions. S.S.K. was supported by NIH award F31HG010818. This research was conducted using the UK Biobank Resource under Application #16549 and was funded by NIH grants R01 GM105857, R01 MH101244, U01 HG009379, and R01 HG006399..
Author information
Authors and Affiliations
Consortia
Contributions
C.M.L. and A.L.P. designed experiments. C.M.L. performed experiments. C.M.L., S.G., P.R.L., S.S.K., N.F. and A.A. analyzed data. C.M.L. and A.L.P. wrote the paper with assistance from S.G., P.R.L. S.S.K., N.F. and A.L.P.
Corresponding authors
Ethics declarations
Competing interests
The authors C.M.L., S.G., P.R.L., S.S.K. and A.L.P. declare no competing interests. N.F. and A.A. and members of the 23andMe research team are employees of 23andMe Inc.
Additional information
Peer review information Nature Communications thanks Matthew Robinson, Doug Speed, and the other, anonymous, reviewer for their contribution to the peer review of this work. Peer reviewer reports are available.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
MárquezLuna, C., Gazal, S., Loh, PR. et al. Incorporating functional priors improves polygenic prediction accuracy in UK Biobank and 23andMe data sets. Nat Commun 12, 6052 (2021). https://doi.org/10.1038/s41467021251719
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467021251719
This article is cited by

Recent advances in polygenic scores: translation, equitability, methods and FAIR tools
Genome Medicine (2024)

Incorporating functional annotation with bilevel continuous shrinkage for polygenic risk prediction
BMC Bioinformatics (2024)

Multiancestry and multitrait genomewide association metaanalyses inform clinical risk prediction for systemic lupus erythematosus
Nature Communications (2023)

Quantifying portable genetic effects and improving crossancestry genetic prediction with GWAS summary statistics
Nature Communications (2023)

Overcoming attenuation bias in regressions using polygenic indices
Nature Communications (2023)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.