An Online Platform for Densely Imputed GWAS Summary Statistics

摘要

Genotype imputation is a necessary step in the Genome-Wide Association Study (GWAS) meta-analyses combining summary statistics of primary GWAS. A survey of publicly available data in the GWAS Catalog and the GRASP database showed that only less than 30% of summary statistics are imputed to the 1000 Genomes (1KG)/Haplotype Reference Consortium (HRC) variant density, with the majority being imputed to the HapMap reference panel. After GWAS meta-analyses are published, genotype imputation is cumbersome and often impossible due to restricted access to individual level data. We implemented a recently published approach, ss-imp to publicly available HapMap-based GWAS summary statistics and we propose to increase their usage by introducing a new web-based platform containing a resource of GWAS summary statistics for multiple phenotypes imputed to 1KG. This removes the necessity for summary statistics re-formatting, parallelizing the processes and preparing the scripts by each researcher in need to use summary statistics imputation. Starting with traits associated with glucose homeostasis, including HOMA-IR, HOMA-B, fasting glucose and insulin, imputed to all-ancestries 1KG, our online resource is constantly growing. The average number of SNPs in HapMap-based summary statistics is 2.5 million before imputation and reaches an average of 8.5 million after imputation to 1KG after stringent quality control. We are in the process of releasing initial set of GWAS summary statistics imputed to the European-ancestry HRC panel. Our platform will advance the research by simplifying the analytical procedures required in both GWAS meta-analyses and other analyses based on summary statistics.