GEDI: Scalable Algorithms for Genotype Error Detection and Imputation

Abstract

Genome-wide association studies generate very large datasets that require scalable analysis algorithms. In this report we describe the GEDI software package, which implements efficient algorithms for performing several common tasks in the analysis of population genotype data, including genotype error detection and correction, imputation of both randomly missing and untyped genotypes, and genotype phasing. Experimental results show that GEDI achieves high accuracy with a runtime scaling linearly with the number of markers and samples. The open source C++ code of GEDI, released under the GNU General Public License, is available for download at http://dna.engr.uconn.edu/software/GEDI/