Effect of genotype and pedigree error on detection of recombination events, sire imputation and haplotype inference using the HSPhase algorithm

Abstract

HSPhase is a fast and accurate algorithm for detection of recombination events, sire imputation and haplotype inference of half-sib families. It can be used on data for half-sib families with as few as 4 individuals in a family. The robustness of this algorithm in relation to genotype and pedigree errors was evaluated. If there were more than 20 half-sibs in a family, the performance of the algorithm with 5% pedigree or genotyping errors was still reliable with the accuracy of phasing and imputation above 0.87. These error rates are above those commonly observed in industry data which indicates the algorithm is sufficiently robust for deployment in real world settings. An R package implementing the method is freely available and includes a function to generate diagnostic plots which are very useful to rapidly identify problems in the dataset.