Abstract
Biobanks often contain several phenotypes relevant to a given disorder, and researchers face complex tradeoffs between shallow phenotypes (high sample size, low specificity and sensitivity) and deep phenotypes (low sample size, high specificity and sensitivity). Here, we study an extreme case: Major Depressive Disorder (MDD) in UK Biobank. Previous studies found that shallow and deep MDD phenotypes have qualitatively distinct genetic architectures, but it remains unclear which are optimal for scientific study or clinical prediction. We propose a new framework to get the best of both worlds by integrating together information across hundreds of MDD-relevant phenotypes. First, we use phenotype imputation to increase sample size for the deepest available MDD phenotype, which dramatically improves GWAS power (increases #loci ~10 fold) and PRS accuracy (increases R2 ~2 fold). Further, we show the genetic architecture of the imputed phenotype remains specific to MDD using genetic correlation, PRS prediction in external clinical cohorts, and a novel PRS-based pleiotropy metric. We also develop a complementary approach to improve specificity of GWAS on shallow MDD phenotypes by adjusting for phenome-wide PCs. Finally, we study phenotype integration at the level of GWAS summary statistics, which can increase GWAS and PRS power but introduces non-MDD-specific signals. Our work provides a simple and scalable recipe to improve genetic studies in large biobanks by combining the sample size of shallow phenotypes with the sensitivity and specificity of deep phenotypes.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
1. We evaluated three alternatives to our phenotype imputation matrix: (1) sex-stratified; (2) adding BMI; and (3) restricting to MTAG phenotypes (Supplementary Figure 2). (1) slightly hurt performance, (2) had little impact, and (3) performed far worse. 2. We significantly improved our presentation of the MTAG results, especially by expanding on our description of its utility in the Discussion and by modifying section titles. 3. We now publicly release our GWAS summary statistics. 4. We clarified/corrected our use of a few key terms, e.g. effective sample size and phenotype integration. 5. We thoroughly evaluated the variances and correlations of the imputed phenotypes (Supplementary Figure 1). 6. We added a major caveat to our Discussion on the potential for phenotype imputation to bias downstream analyses that others might perform, as well as potential solutions for these future directions. 7. We formally show that the increase in GWAS power from phenotype integration is statistically significant (Supplementary Figure 3). 8. We expanded our Discussion and added relevant references to place our work in context of prior efforts to improve power for MDD GWAS, including proxy GWAS (GWAX) and combining endorsements of multiple depression measures. 9. We extensively characterized PRS Pleiotropy as a function of sample size (Extended Data Figure 5). Most importantly, our results confirm that the conclusions in our paper are robust to sample size differences between our PRS. More broadly, our results characterize the complex interplay between sample size, p-value threshold, and the number of SNPs in a PRS (Supplementary Figures 10-11, Extended Data Figure 6). 10. We added two key references to prior work on MTAG; a theoretical result that complements our empirical results on power-specificity tradeoffs, and a smaller-scale observation consistent with our result that MTAG inflates genetic correlation. 11. We corrected a small but important overstatement in our previous submission on the relationship between portability and biological causality.
