Benchmarking 24 combinations of genotype pre-phasing and imputation software for SNP arrays in pigs

Evaluated 24 combinations of imputation software for pigs. Beagle-Minimac combination provided the best imputation accuracy. Beagle-Beagle combination stands out for convenience. Eagle-pbwt combination showed excellent performance in resource efficiency. Genotype imputation is essential for increasing marker density and maximizing the utility of existing SNP array data in animal breeding. Although a wide range of software is available for genotype imputation, a comprehensive benchmark in pigs is still lacking. In this study, we benchmarked 24 combinations of genotype imputation software for SNP arrays in pigs, comprising six independent pre-phasing software (fastPHASE, MaCH, BIMBAM, Eagle, SHAPEIT, Beagle) and four distinct imputation software (pbwt, Minimac, IMPUTE, Beagle), using 1,602 whole-genome sequencing (WGS) pigs from a multibreed pig genomics reference panel (PGRP) in PigGTEx. Our results indicated that the combination of Beagle for pre-phasing and Minimac for imputation achieves the highest imputation accuracy with a concordance of 0.983, especially for low-frequency SNPs (MAF<0.05). Finally, we proposed three recommended strategies: i) the combination of Beagle and Minimac is optimal for achieving the highest accuracy; ii) the combination of Beagle and Beagle is recognized for its convenience and relatively high accuracy despite it being memory-intensive; iii) the combination of Eagle and pbwt is feasible for its minimal computational cost with relatively high accuracy. This study provides valuable insights for implementing genotype imputation for pig SNP arrays toward sequence data and offers a basis for applications in livestock and poultry breeding.