Figure 4
From: Recent natural selection conferred protection against schizophrenia by non-antagonistic pleiotropy
Directionality of RNS enrichment in schizophrenia considering the derived alleles emerged during human evolution. (A) Average log10(ORSCZ) of RNS markers (orange) and the remaining SNPs (grey), referred to the derived alleles that emerged during human evolution, across different SCZ GWAS thresholds. 95% confidence intervals of the means are displayed. (B) Proportion of SNPs within RNS markers and the remaining SNPs with a protective or risk effect of the derived allele, across different schizophrenia GWAS thresholds. (C) PGSSCZ predictions in the case–control target sample (NSCZ = 1927; NHC = 1561) stratified by RNS. Schizophrenia GWAS summary statistics were subsetted based on the relationship of the SNP to RNS and the risk or protective effect of the derived allele as follows: (i) RNS markers (top 5% of the SNPs with the highest probability of being subject to RNS) whose derived alleles arisen through human evolution confer protection to schizophrenia (derived-ORSCZ < 1; light brown), (ii) RNS markers (top 5% of the most selected SNPs) whose derived alleles arisen through human evolution confer risk to schizophrenia (derived-ORSCZ > 1; dark brown), (iii) SNPs not subject to RNS (from the remaining 95% of SNPs) whose derived alleles arisen through human evolution confer protection to schizophrenia (derived-ORSCZ < 1; light grey), (iv) SNPs not subject to RNS (from the remaining 95% of SNPs) whose derived alleles arisen through human evolution confer risk to schizophrenia (derived-ORSCZ > 1; dark grey). Explained variance attributable to PGS was calculated as the increase in Nagelkerke’s pseudo-R2 between a linear model with and without the PGS variable. P values were obtained from the binomial logistic regression of SCZ phenotype on PGS, accounting for LD and including sex, age, and 10 MDS ancestry components as covariates. Significant PGS predictions after FDR correction (pFDR < 0.05) are marked with an asterisk. (D) Distribution of variance explained (% pseudo-R2) from stratified PGSSCZ predictions based on the abovementioned subsets, using a similar number of variants (NSNPs = 1000 from all available SNPs (PSCZ < 1)) for each analysis. Wilcoxon test was used to compare 1000 pseudo-R2 distributions from PGS predictions. See Supplementary Data 7 for detailed results.
