Detecting CNVs in Porcine Genotypes

All 120 samples were genotyped using a Porcine SNP60K BeadChip, and the raw data were extracted using GenomeStudio (Version 1.9.4, Illumina, Inc.). We used a strict quality control threshold (call rate > 98%, call frequency > 90%) to minimize the false-positive rate and eliminate low-quality samples. After quality control, no samples were excluded from CNV detection. The PennCNV software using a hidden Markov model (HMM), which allows detection of CNVs from Illumina or Affymetrix SNP chip data, was employed for CNV identification. PennCNV integrates various sets of information, including the total signal intensity of the log R Ratio (LRR), the distance between neighboring SNPs, the B allele frequency (BAF), the population frequency of the B allele (PFB) of SNPs, pedigree information, and the–gc model to adjust for “genomic waves” [11 (link), 43 , 65 (link), 66 (link)]. We could not obtain pedigree information for many of the samples because the samples were obtained from multiple locations across Congjiang County, China. Thus, pedigree information was not incorporated into the analysis. Moreover, the CNVs were aggregated using CNVRuler software [67 (link)].