Extended Data Fig. 2: Forest plot summaries of quantitative and binary trait associations with GDF15 LOF and C211G variants.
From: Identification and characterization of human GDF15 knockouts
Forest plots, with per standard deviation (S.D.) units and 95% confidence intervals, of traits for GDF15 LOF and C211G carriers from 75,018 samples (158 heterozygotes, 5 knockouts) of PGR. (a) Quantitative traits. (b) Binary traits. Genotype counts for cases and controls are given as HomRR|HetRA|HomAA; R=reference allele; A=alternate allele. P-values are generated using whole genome regression models adjusting for age, sex, age*sex, age2 and top 10 genetic principal components (PCs). Additionally, Firth correction was used for binary traits. Quantitative effect estimates, per standard deviation (S.D.) units, are generated by standard linear regression adjusting for age, sex, age*sex, age2 and top 10 genetic PCs. No phenotype had FDR-5% p-value < 0.05 after applying Benjamini-Hochberg correction.
