Goncalo Abecasis
Professor
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (3)
Through an analysis of 2,602 genome-wide association studies (GWAS) across 830 human traits, we find that most (56% of) well-studied traits have at least two published GWAS, and many (29%) have at least five. We show that the lack of an established approach for adjudicating variant association estimates across multiple published studies can lead to uncertainty and invalid inferences: using all associations ever published for a trait increases true positives (by 12%) but also false positives (by 55%) relative to using associations from the largest published GWAS for the trait. We employ a “bottom-line” procedure for meta-analyzing published GWAS while inferring and accounting for sample overlap, which identifies a more accurate and comprehensive list of associations relative to existing approaches. Five commonly used bioinformatic methods for post-GWAS analyses produce reliable results when applied to the bottom-line associations. We present these results for 1,281 human complex traits, including 1,839 single-ancestry and 576 trans-ancestry analyses, for browsing or download via the NHGRI Association to Function Knowledge Portal. This resource of “consensus” GWAS results is intended to increase replicability, reuse, and interpretation of GWAS and downstream analyses.
Year:
2026
Meta-analysis of gene-based tests using single variant summary statistics is a powerful strategy for associating genes with disease. However, current approaches require sharing the covariance matrix between variants for each study and trait of interest. For large-scale studies with many phenotypes, these matrices can be cumbersome to calculate, store, and share. To address this challenge, we present REMETA, an efficient tool for meta-analysis of gene-based tests. REMETA uses a single sparse covariance reference file per study that is rescaled for each phenotype using single variant summary statistics. We develop methods to apply REMETA to binary traits with case-control imbalance, and estimate allele frequencies, genotype counts and effect sizes of burden tests. We demonstrate the performance and advantages of our approach via meta-analysis of 5 traits in 469,376 samples in UK Biobank. The open-source REMETA software tools and framework will facilitate meta-analysis across large scale exome sequencing studies from diverse studies that cannot easily be brought together.
Year:
2024
Whole-genome sequencing (WGS), whole-exome sequencing (WES) and array genotyping with imputation (IMP) are common strategies for assessing genetic variation and its association with medically relevant phenotypes. To date, there has been no systematic empirical assessment of the yield of these approaches when applied to hundreds of thousands of samples to enable the discovery of complex trait genetic signals. Using data for 100 complex traits from 149,195 individuals in the UK Biobank, we systematically compare the relative yield of these strategies in genetic association studies. We find that WGS and WES combined with arrays and imputation (WES + IMP) have the largest association yield. Although WGS results in an approximately fivefold increase in the total number of assayed variants over WES + IMP, the number of detected signals differed by only 1% for both single-variant and gene-based association analyses. Given that WES + IMP typically results in savings of lab and computational time and resources expended per sample, we evaluate the potential benefits of applying WES + IMP to larger samples. When we extend our WES + IMP analyses to 468,169 UK Biobank individuals, we observe an approximately fourfold increase in association signals with the threefold increase in sample size. We conclude that prioritizing WES + IMP and large sample sizes rather than contemporary short-read WGS alternatives will maximize the number of discoveries in genetic association studies.
Year:
2024
Collaborators (5)
Timothy Thornton
Professor
Emory University
Olivier Delaneau
Assistant Professor
University of Lausanne
Jonathan Marchini
Head of Statistical Genetics and Machine Learning
Regeneron Pharmaceuticals
Manuel Ferreira
QIMR Berghofer Medical Research Institute
Michael Boehnke
Professor
University of Michigan
Make your next move.
Prepare your application
Review drafts before you send them.