GENOMIC VARIATION IN LIVESTOCK USING DENSE SNP CHIP DATA

Part I describes two possible approaches to investigate Mexican chicken genetic variation, using selective sweeps and Copy Number Variants (CNV). CNVs are genomic polymorphisms that influence phenotypic expression and are an important source of genetic variation in populations. The aim of the first study here presented was to characterize the genetic variability of the Mexican chicken’s population and to disclose any underlying population structure. A total of 213 chickens were sampled in different rural production units located in 25 states of México. Genotypes were obtained using the Affymetrix Axiom® 600K Chicken Genotyping Array. The Identity by Descent (IBD) and the Principal Components Analysis (PCA) were performed by SVS software on pruned SNPs. Analyses done with ADMIXTURE identified three ancestors and determined, for each individual, the proportion of the genetic contribution from each of the three ancestors. The results of the Neighbor-Joining (NJ) analysis were consistent with those obtained by the PCA. All methods used in this study did not allow a classification of Mexican chicken in distinct genetic groups. A total of 3,059 Run of homozygosity (ROH) were identified and, being mainly short in length (< 4 Mb), these regions are indicative of a low inbreeding level in the population. Finally, findings from the ROH analysis indicated the presence of natural selective pressure in the population of Mexican chicken. In the second study we used CNVs to investigate genetic variability in the Mexican Creole chicken and to relate this variation to the available gene annotation. The Hidden Markov Model of the PennCNV software detected a total of 1,924 CNVs in the chicken genome of 256 individuals. Input data were LOGR Ratio and B allele frequency obtained with the Axiom® Genome-Wide Chicken Genotyping Array (Affymetrix). The mapped CNVs comprised 1,538 gains and 386 losses resulting, at population level, in 1,216 CNV regions (CNVRs), of which 959 gains, 226 losses and 31 complexes (i.e. containing both losses and gains). The CNVRs covered a total of 47 Mb of the whole genome sequence length, corresponding to 5.12 % of the chicken galGal4 autosome assembly. This study allowed a deep insight into the structural variation in the genome of unselected Mexican chicken population, which up to now has not been genetically characterized. The genomic study disclosed that the population, even if presenting extreme morphological variation, couldn’t be organized in differentiated genetic subpopulations. Finally, this study provided a chicken CNV map based on the 600K SNP chip array, jointly with a genome-wide gene copy number estimates in a native, unselected for more than 500 years, chicken population. Genetic variation can be caused by adaptive evolutionary changes and by artificial selection. The genetic makeup of populations is the result of a long-term process of selection and adaptation to specific environments and ecosystems. The two studies here presented indicate that the Mexican chicken clearly appear to be a unique Creole chicken population that was not subjected to a specific directional selection. Results provide a genetic knowledge that can be used as a basis for the genetic management of a unique genetic resource. Industry is likely envisaging to use the female native populations mating them with selected males to increase the productivity and the economic revenue of family farming agriculture, which is a large reality of United States of México. • Part II describes a CNV scan and a population analysis of turkey populations coming from different countries. The domesticated turkey was brought to Europe in late 1500 by Spanish conquerors from Central America, likely from Mexico. The evolution of the Mexican turkey population occurred as such independently for more than 500 years from the European ones and the commercial hybrids. This study investigates the genomic diversity of several turkey populations using CNVs as source of variation. A total of 116 individuals from 6 Italian breeds (Colle Euganei, Bronzato Comune Italiano, Parma e Piacenza, Brianzolo, Nero d’Italia and Ermellinato di Rovigo), 7 Narragansett, 38 commercial hybrids and 31 Mexican turkeys, were processed with the Affymetrix 600K SNP turkey array. The CNV calling was performed with the HMM of PennCNV software. CNV were summarized into CNV regions (CNVRs) at population level using BEDTools. Variability among populations has been addressed by hierarchical clustering (pvclust R package) and by principal component analysis (PCA). A total of 2,987 CNV were identified covering 4.65% of the autosomes of the Turkey_5.0/melGal5 assembly. The CNVRs including at least 2 individuals were 362, 189 gains, 116 losses and 57 complexes. Among these regions the 51% contain genes. This study is the first CNV mapping of turkey population using 600K SNP chip. CNVs clustered the individuals according to population and their geographical origin. CNVs are also known to be indicators of adaptation, as some researches are suggesting investigating different species. The outcomes of this are likely reflecting the human action on domestication of domesticated turkey after its introduction into Europe and the directional selection occurring in the last 40 years to produce a fast-growing heavy bird. • Part III describes the CNV mapping in the Valdostana Red Pied (VRP) cattle breed, an autochthonous Italian dual-purpose cattle population reared in the Alps, and the comparison with the CNV maps detected in previous studies in the Italian Brown Swiss (IBS) and in the Mexican Holstein (HOL). Many studies have focused on identifying CNVs within and between human and livestock populations alike, but only few have explored population-genetic properties in cattle based on CNVs derived from a high-density SNP array. In this study in cattle we report a high-resolution CNV scan, using the Illumina 777k BovineHD Beadchip, for VRP, a population that did not undergo strong selection for production traits. After stringent quality control and filtering, CNVs were called across 108 bulls using the PennCNV software. A total of 6,784 CNVs were identified, summarized to 1,723 CNV regions (CNVRs) on 29 autosomes covering a total of ~59 Mb of the UMD3.1 assembly. Among the mapped CNVRs, there were 812 losses, 832 gains and 79 complexes. A total of 171 CNVRs were common to VRP, IBS and HOL. Between VRP and IBS, 474 regions overlapped, while only 313 were in common between VRP and HOL, indicating a more similar genetic structure among populations with common origins, i.e. the Alps. The clustering and admixture analyses showed a clear separation of the three breeds into three distinct clusters. In order to describe the distribution of CNVs within and among breeds we used the pair VST statistic. We considered only the CNVRs shared by more than 5 individuals within breed. We identified unique and highly differentiated CNVs (n=33), some of which could be due to specific breed selection and adaptation. Genes and QTL within these regions were also characterized adding evidence to the relationship between CNV and adaptation.