He, J.-C. | bioRxiv Science

bioRxiv2024

UK BioCoin: Swift Trait-Specific Summary Statistics Regression for UK Biobank

Summary statistics derived from large-scale biobanks facilitate the sharing of genetic discoveries while minimizing the risk of compromising individual-level data privacy. However, these summary statistics, such as those from the UK Biobank (UKB) provided by Neales lab, are often adjusted by a fixed set of covariates to all traits (12 covariates including 10 PCs, sex and age), preventing the exploration of trait-specific summary statistics. In this study, we present a novel computational device UK BioCoin (UKC), which is designed to provide an efficient framework for trait-specific adjustment for covariates. Without requiring access to individual-level data from UKB, UKC leverages summary statistics regression technique and resources from UKB (289 GB of 199 phenotypes and 10 million SNPs), to enable the generation of GWAS summary statistics adjusted by user-specified covariates. Through a comprehensive analysis of height under trait-specific adjustments, we demonstrate that the GWAS summary statistics generated by UKC closely mirror those generated from individual-level UKB GWAS ({rho} [≥] 0.99 for effect sizes and{rho} [≥] 0.99 for p-values). Furthermore, we demonstrate the results for GWAS, SNP-heritability estimation, polygenic score, and Mendelian randomization, after various trait-specific covariate adjustments as allowed by UKC, indicating UKC a platform that harnesses in-depth exploration for researchers lacking access to UKB. The whole framework of UKC is portable for other biobank, as demonstrated in Westlake Biobank, which can equivalently be converted to a UKC-like" platform and promote data sharing. UKC has its computational engine fully optimized, and the computational efficiency of UKC is about 70 times faster than that of UKB. We package UKC as a Docker image of 20 GB (https://github.com/Ttttt47/UKBioCoin), which can be easily deployed on an average computer (e.g. laptop). One sentence summaryWe develop UK BioCoin (UKC), which allows fine-tuning of covariates for each UK Biobank trait but does not relay on UK Biobank individual-level data. It will change the current landscape of GWAS and reshape its downstream analyses.

genetics↗

bioRxiv2023

Rapid and accurate multi-phenotype imputation for millions of individuals

Deep phenotyping can enhance the power of genetic analysis, including genome-wide association studies (GWAS), but the occurrence of missing phenotypes compromises the potential of such resources. Although many phenotypic imputation methods have been developed, the accurate imputation of millions of individuals remains extremely challenging. In the present study, we developed a novel multi-phenotype imputation method based on mixed fast random forest (PIXANT) by leveraging efficient machine learning (ML)-based algorithms. We demonstrate that PIXANT runtime is faster and computer memory usage is less than that of other state-of-the-art methods when applied to the UK Biobank (UKB) data, suggesting that PIXANT is scalable to cohorts with millions of individuals. Our simulations with hundreds of individuals showed that PIXANT accuracy was superior to or comparable to the accuracy of the most advanced methods available. PIXANT was used to impute 425 phenotypes for the UKB data of 277,301 unrelated White British citizens. When GWAS was subsequently performed on the imputed phenotypes, 18.4% more GWAS loci were identified than before imputation (8,710 vs 7,355). The increased statistical power of GWAS identified novel positional candidate genes affecting heart rate, such as RNF220, SCN10A, and RGS6, suggesting that the use of imputed phenotype data from a large cohort may lead to the discovery of novel genes for complex traits.

bioinformatics↗