Fig. 4
From: The effects of sample size on population genomic analyses – implications for the tests of neutrality
Influence of sample size on the estimation of Fu-Li D FL . a The difference between the total number of mutations (η) and singleton mutations (η s ) estimated for neutral (η S and η Ss ) and constrained (η N and η Ns ) sites using various numbers of human exomes. b Scatter plot shows Fu and Li D estimated for nonsynonymous (D FLN ) and synonymous sites (D FLS ) using different sample sizes. The relationships were significant at the 1 % level. c Relationship between the magnitude of selection pressure on protein-coding genes (dN/dS) and the difference (δ DFL ) in Fu-Li D FL estimated using large (N = 512) and small (N = 16) number of exomes. The plot shows the comparative patterns of the relationships observed for neutral (δ DFLS ) and constrained (δ DFLN ) sites. Each data point indicates the δ DFLS or δ DFLN estimated using the genes belong to a selection intensity category (eg. dN/dS = 0.1-0.2). The correlation involving constrained sites was significant (P < 0.01) but that of neutral sites was not (P = 0.14)