Mining whole genome sequencing data from more than 600 pigs of different breeds to identify variants in coat colour genes

Coat colour has been one of the first traits that was selected over the domestication process in all livestock species. This selection acted indirectly at the genome level where variants have been fixed or almost fixed in many breeds. The domestication process of the pig started from Asian, Middle East and European wild boar ancestors and spread mainly in China and Europe. Therefore, subsequently coat colour affecting genes have followed two main independent selection processes creating a quite high level of variability. Many genes are known to control pigmentation in mammals, including genes (i) affecting melanocytes morphology, development and migration, (ii) coding for melanogenic enzymes and (iii) affecting the melanosomal structure and function. In this study, we selected 25 coat colour genes, involved in all relevant biological processes related to pigmentation, and investigated their variability by mining more than 600 whole genome sequencing (WGS) datasets produced from 35 Sus scrofa breeds/populations and related species (S. cebifrons, S. celebensis and S. verrucosus). About 500 WGS datasets from cosmopolitan and European and Asian autochthonous pig breeds were retrieved from the European Nucleotide Archive. Additional 100 WGS data were produced from three Italian heavy pig breeds. Reads were mapped to the reference genome with BWA and polymorphisms were detected with Bcftools in the selected genes, considering a region spanning 5 kb upstream and 5 kb downstream the corresponding gene coordinates. Variants were annotated using the Variant Effect Predictor tool and SIFT was used for predicting the impact on the protein function. In total, we analysed more than 10 Terabyte of sequencing data. A total of more than 150 k variants were identified, including about 7 k insertion/deletions (~5%). About 90% of all these variants were located outside the protein coding region (~80% in introns, ~10% in UTRs and ~10% flanking UTRs). Few variants might putatively affect the protein structure: about 600 and 200 missense variants were predicted by SIFT as tolerated and deleterious, respectively; about 30 variants determined stop codons or frameshifts. Allele frequencies for all variants were estimated across the different breeds. Variants were compared to establish their relationships with the known phenotypes and to confirm their putative role as causative variants. Variants were used to infer the coat colour of the pigs from which WGS data were produced and allocate animals to their breeds. Moreover, many variants were also used to verify the breed of origin of pork products.