2011•PubMedOpen access
The effect of reference panels and software tools on genotype imputation.
Kwangsik Nho, Li Shen, Sungeun Kim, Shanker Swaminathan, Shannon L. Risacher, Andrew J. Saykin
Open full text 8 citations
Abstract
Genotype imputation is increasingly employed in genome-wide association studies, particularly for integrative and cross-platform analysis. Several imputation algorithms use reference panels with a larger set of genotyped markers to infer genotypes at ungenotyped marker locations. Our objective was to assess which method and reference panel was more accurate when carrying out imputation. We investigated the influence of choice of two most popular imputation methods, IMPUTE and MACH, on two reference panels from the HapMap and the 1000 Genomes Project. Our results indicated that for the HapMap, MACH consistently yielded more accurate imputation results than IMPUTE, while for the 1000 Genomes Project, IMPUTE performed slightly better. The best imputation results were achieved by IMPUTE with the combined reference panel (HapMap + 1000 Genomes Project). IMPUTE with the combined reference panel is a promising strategy for genotype imputation, which should facilitate fine-mapping for discovery as well as known disease-associated candidate regions.