We combined our sequenced samples with a further 185 genomic sequences from the NCBI SRA database. This combined dataset contained data from a total of 199 Bos individuals, including 106 taurine, 32 indicine, 39 gayal, 10 banteng, and 12 wisent (European bison, Bison bonasus; supplementary table S1, Supplementary Material online). All clean reads were mapped to the assembled gayal reference genome using BWA-MEM v0.7.12 (Li 2013 (link)). PCR duplicate reads were sorted and removed using samtools v1.3.1 (Li 2011 (link)). Next, RealignerTargetCreator and IndelRealigner in the Genome Analysis Toolkit (GATK) v3.7.0 (DePristo et al. 2011 (link)) were employed for local realignment around indels. SNPs were identified using UnifiedGenotyper from GATK package with default parameters. Finally, we applied the following hard filtering criteria to all SNPs: 1) QUAL < 30; 2) QualByDepth < 2.0; 3) RMSMappingQuality < 40.0; 4) MappingQualityRankSumTest < −12.5; 5) ReadPosRankSumTest < −8.0; and 6) HaplotypeScore > 13.0. Genotypes were imputed and phased with BEAGLE v4.1 (Browning and Browning 2007 (link)).