Long-range Allele Frequency Differences for Selection Signature Analysis

Allele frequency change is the fundamental change due to genetic selection. The availability of genome-wide SNP markers allows the study of allele frequency differences (AFD) between populations at the whole genome level. However, single-locus AFD are highly variable and difficult to interpret. Long-range heterozygosity measures in sliding windows have been proposed for selection signature analysis. In this research, we developed two long-range AFD measures for selection signature analysis, AFD and standardized AFD in sliding windows of SNP markers. For each chromosome, the standardized AFD may use within-line or accross-line mean and standard deviation of the AFD values of the entire chromosome, with the within-line standardization being more conservative. With these long-range AFD measures, what looked trendless single-locus AFD would show a trend of genome differences for chromosome regions that had been subjected to selection. The appropriate window size is dependent on the density of SNP coverage of the genome. Small window sizes generally have larger variations in the AFD and standardized AFD values and are less trendy, but the peak AFD and standardized AFD values are closer to the single-locus peak locations. In contrast, larger window sizes have the opposite effects to smaller window sizes. These two long-range AFD measures compliment existing long-range hterozygosity measures for selection signature analysis, particularly in cases of ‘p-q sweep’, where genetic selection changed the initial allele frequency of ‘p’ into ‘q’ and changed the initial ‘q’ into ‘p’ so that heterozygosity remained unchanged even though selection substantially changed the allele frequencies. The long-range AFD measures, as well as the existing long-range heterozygosity measures, are computationally efficient and are suitable for genome-wide screening of selection signatures.