Whole genome resequencing and SNP detection
Genomic DNA was isolated and purified from the ethanol stored fin clips using Macherey-Nagel nucleomag tissue kit, following the manufacturer’s protocol. Paired end, PCR-free 150-bp insert libraries were then prepared for whole genome sequencing using the DNBSeqTM platform by BGI-Hongkong, generating an average of 40 million cleaned reads per individual (min = 33.4 million, max = 41.8 million, equating to an average depth of coverage of 10X). The clean paired-end reads were aligned to the threespine stickleback genome assembly v5 with BOWTIE2 (version 2.4.1) using default parameter settings , and sorted and indexed using SAMTOOLS (version 1.10) . Variants were discovered using the short variant discovery pipeline of GATK . SNPs and indels were quality filtered according to GATKs best practices guidelines, using the following hard filters: QualByDepth > 2.0, FisherStrand bias < 60.0, RMSMappingQuality < 40.0, MappingQualityRankSumTest < -12.5, ReadPosRankSumTest < -8.0, StrandOddsRatio > 3.0, variant quality score < 30.0. For all following analyses, we removed mitochondrial variants, indels and multiallelic variants, as well as variants identified on either of the sex chromosomes . We then filtered the remaining autosomal SNPs for genotype depth less than six or greater than 100; minor allele counts less than four; missingness less than 20%. The sex of individuals was confirmed using the proportion of reads with depth greater than eight mapped to the X vs Y chromosome .