Evaluation of the BEAGLE Haplotype Reconstruction Algorithm: Using Half Sib Families to Evaluate the Accuracy of Haplotype Reconstruction

Abstract

The BEAGLE program is frequently used for haplotype reconstruction using dense molecular marker genotype data in animal, plant or human populations. In this study, we evaluate the accuracy of BEAGLE for phasing with a population that consists of large half-sib families. The half-sib structure allows the accurate detection of phase thus making it easier to detect problems in more general population based algorithms such as BEAGLE. We show that the main problem in the haplotypes inferred by BEAGLE is the occurrence of switch errors where the parental origin of haplotypes is incorrectly swapped. This occurs especially in the large chromosomes and often in ~50% of individuals. Understanding the issue will allow better decision making about further analyses that relies on the haplotype origin of markers.