DNA Sequencing and SNP Calling
Three to five individuals were randomly selected from each population whose phenotype was measured to extract DNA. Total genomic DNA was extracted using a modified cetyltrimethylammonium bromide (CTAB) method (Doyle & Doyle, 1987). Illumina Xten from Biomarker Technologies, Inc. (Beijing, China) was used for genomic library generation and sequencing with 2 × 150 bp paired reads. Furthermore, the raw reads of Semiaquilegia adoxoide s (SRR437677) were downloaded from the NCBI SRA database (http://www.ncbi.nlm.nih.gov/sra) to be used as an outgroup. To obtain high-quality genomes, all the reads were assessed using FastQC (Andrew, 2010) and filtered as follows: reads with adapters and reads with more than a 10% N content or more than 50% low-quality bases (quality value of less than 10) were removed. Low-quality reads were removed using NGStoolkit (Mulcare, 2004). The clean reads of each sample were mapped to the genome sequence of A. coerulea downloaded from the previous study of Filiault et al. (Filiault et al., 2018) using BWA v.0.7.12 with default parameters (H. Li & Durbin, 2009). SAMtools v.0.1.18 was used to sort reads (H. Li et al., 2009). The HaplotypeCaller, GenotypeGVCFs and CombineGVCFs modules in GATK v.4.1.8.0 were used to produce accurate SNP calls (McKenna et al., 2010). To improve the quality of SNPs, the VariantFiltration module in GATK v4.1.8.0 was used for filtration with the following parameters: –filter-name FilterQual –filter-expression ”QUAL < 30.0” –filter-name FilterQD –filter-expression ”QD < 2.0” –filter-name FilterMQ –filter-expression ”MQ < 40.0” –filter-name FilterFS –filter-expression ”FS > 60.0” -window 5 -cluster 2. Next, VCFtools v0.1.13 (Danecek et al., 2011) was used to remove variants that 1) showed a minor allele frequency (MAF) of 0.02 or less, 2) were not balletic variants, 3) showed a sequencing depth of less than 5, and 4) showed a missing rate exceeding 0.5. The filtered data were used for subsequent analysis. After filtering, each SNP was annotated using SnpEff v.5.0 (Cingolani et al., 2012).