Statistical analyses were performed to search for signatures of selection using the 274 genotypes from generation 11 (139 from the line with high VE of LS and 135 from the line with low VE of LS) and 90 genotypes from the base population. The 93 genotypes from generation 13 were kept for the validation analysis. Three methods were used to identify the patterns of signatures of selection (Fig. 1): (a) detection of runs of homozygosity (ROH), (b) quantification of the variation in LD patterns (VarLD), and (c) estimation of the fixation index (FST).

Methods of identifying patterns of signatures of selection. a Runs of homozygosity (ROH). From left to right: the number of consecutive homozygous SNPs increases, generating a genomic region where the individual is homozygous at all sites, i.e. a ROH. b Fixation index (FST). From left to right: allele frequencies of individuals in the population change until it differentiates into two different sub-populations (FST = 1). c Variation in linkage disequilibrium (VarLD), which searches for differences in linkage disequilibrium (LD) patterns between populations. From left to right: an advantageous allele (red star) can modify the LD in a population because of the selective sweep containing the SNPs surrounding it, i.e. the haplotype of this advantageous allele (Haplotype 2)

A ROH is a region of the genome that displays a local reduction of genetic variation, i.e. a genomic region for which the individual is homozygous at all sites, which indicates the presence of a locus that is affected by selection (Fig. 1a) [20]. Using an algorithm implemented in PLINK v1.9 [17], we identified the ROH in all the individuals from the base population and from the lines with high and low VE of LS of generation 11. The parameters were set according to Ceballos et al. [21]. This algorithm searches for stretches of consecutive homozygous SNPs on each chromosome using sliding windows of 500 kb that contain around 50 SNPs. SNPs with missing calls and more than one heterozygous SNP were not allowed in a window. The proportion of the overlapping windows that must be called homozygous to define any given SNP as in a homozygous segment was set to 0.05%. Two SNPs separated by more than 1 Mb belonged to two different homozygous segments. A homozygous segment was considered as a ROH if the number of consecutive SNPs exceeded 50 and the SNP density was higher than one SNP per 30 kb. A ROH must be a consensus genomic region in the selected animals to be a candidate signature of selection, i.e. it had to be identified in 50% of the animals in the line with low VE of LS (65), and in 50% of the animals in the line with high VE of LS (70).

The fixation index (FST) was used to estimate the differences in allele frequencies between the lines with high and low VE of LS (Fig. 1b). The FST was calculated using Weir and Cockerham’s pairwise estimator method [22], implemented in the VCFtools v1.16 software [23]. The FST values were estimated in 500-kb sliding windows with a step size of 250 kb. Windows with less than ten SNPs were excluded from the analysis. FST values were weighted to take differences in sample sizes between populations into account (for further details see Weir and Cockerham [22]). Relevant FST windows were those with a weighted FST value equal or above the weighted FST value in the 99.9th percentile of the distribution for all the genomes. MAF was calculated in the base population and the lines with high and low VE of LS for the relevant FST windows. Those that showed divergent changes in MAF between the rabbit lines relative to the base population were considered to be putative signatures of selection. These windows were considered as resulting from an effect of genetic drift if the MAF between the lines with high and low VE of LS at generation 11 displayed the same change relative to the base population (increase or decrease) or if one of the lines did not show any change (i.e. had a MAF equal to that in the base population).

We used the VarLD software [24] to evaluate the magnitude of the differences in LD patterns (Fig. 1c) between two populations. We analysed the pairwise comparison of the three populations: base population with the line with high VE of LS (Base-High), base population with the line with low VE of LS (Base-Low), and between the lines with high and low VE of LS (High–Low). Sliding windows of 50 SNPs with a step size of one SNP were used to calculate the correlation matrix of each population per chromosome. The program computed the r2 metric for each pair of SNPs to determine the strength of LD in each window. The difference between the eigenvalues of the correlation matrices of both populations determined the VarLD score, which was standardized by the mean and the standard deviation of all the scores along each chromosome. A genomic window was relevant when its standardized VarLD scores were equal or above the standardized VarLD score in the 99.9th percentile distribution for all the genomes. The relevant windows identified in both the Base-High and Base-Low comparisons were considered as putative signatures of selection and the relevant windows identified only in the High–Low comparison were considered as resulting from the effect of gene drift.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.