Completed on 4 Dec 2017 by Hubert Pausch. Sourced from https://www.biorxiv.org/content/early/2017/11/30/227157.
Login to endorse this review.
Whalen et al. review the performance and use of HMM-based imputation methods in animal breeding populations. Their initiative to review existing methods is highly appreciated because HMM-based imputation methods have been widely used to infer haplotypes and impute microarray and sequence variant genotypes in many species. Because the preprint does not include a *single* reference to a study that evaluated the performance of HMM-based methods for genotype imputation in livestock populations, one would certainly conclude from reading the pre-print that HMM-based approaches are barely considered to impute missing genotypes in animal populations.
However, most of the methods reviewed in the preprint have been widely used to infer haplotypes and impute missing genotypes for both microarray and sequence derived variants in animal populations (e.g., https://www.ncbi.nlm.nih.go... , https://www.ncbi.nlm.nih.go... , https://www.ncbi.nlm.nih.go... , https://www.ncbi.nlm.nih.go... ). In the present preprint, Whalen et al. show that pre-phasing-based imputation is both highly accurate and computationally efficient - a similar conclusion was drawn using genome-wide marker data in a cattle population five years ago . Interestingly, the preprint shows that combining Eagle and Minimac gives very high accuracies in both simulated and real data. These tools were used to impute sequence variant genotypes in large cattle populations at high accuracy considering various reference populations (https://www.ncbi.nlm.nih.go... , https://www.ncbi.nlm.nih.go... ).
While I appreciate that the conclusions drawn by Whalen et al. are both timely and relevant for the livestock genomics community, they would be more compelling if they were backed up by evidence that is widely available from literature.
COI statement: Please note that I'm the corresponding author of some of the above mentioned publications [4-6].
Hubert Pausch, ETH Zürich