Comprehensive Benchmarking of Genotype Imputation Tools Using a Large-Scale Chinese Reference Panel

Genome-wide association studies (GWAS) remain as one of the most essential and potent strategies for identifying genetic markers associated with common human diseases. Gene arrays, which are widely used in GWAS, target predesigned known genomic mutations to enhance clinical diagnostics. The approach of genotype imputation which leverages reference panels has shown great promise in extrapolating additional meaningful variants from existing data. Therefore, accurately broadening the spectrum of mutations beyond those included in gene arrays can unlock significant potential for cost-effective and advanced GWAS analysis. In this study, we focus on the four most popular imputation tools currently to assess their performances based on a large-scale Chinese reference panel. Benchmarking results demonstrate that Minimac4 is the best imputation method, which achieved higher accuracy with more well-imputed loci and maintained lower computational consumption. Moreover, we further revealed the minimum sample size required in the reference panel and the mutations involved in arrays that ensure the acquisition of acceptable imputed call sets. We believe this study can provide practical significance guidelines for the research community to select the suitable imputation tool via an existing reference panel, or construct their own reference panel and genotype array in the most effective way.