The left panel of Figure 1 shows that, as expected, more SNPs were selected as the P-value threshold became less stringent. The proportion of selected SNPs was almost identical to the P-value threshold, indicating that the P-values obtained in the association test were not inflated. SNPs with P ≤ 0.001 explained only 0.68% of the phenotypic variance (the median value based on 50 random partitions, the right panel of Figure 1). As the P-value threshold became less stringent, the explanatory value of the sets of SNPs increased and then plateaued; about 17.6% of the phenotypic variance could be explained by SNPs with P ≤ 0.1. This indicates that most of the chip-heritability can be attributed to SNPs with P ≤ 0.1. We further checked the linear relationship between the proportion of SNPs and their explained variance based on 50 random partitions; the obtained R2 = 19.8%(P = 1.62 × 10−14), reflecting a highly significant association between the proportion of SNPs and their explained variance.
