Fig. 2

Case study 1: power under an ordinal multinomial model. This example shows the power to detect a single causal SNP in UK Biobank data with four outcome categories for disease status. Using an ordinal multinomial simulation model and the OpenMendel module for ordinal trait regression [13], we assume a single SNP as a fixed effect and control for sex and standardized age. The figure compares analysis results for three SNPs of varying MAF over 1000 simulation replicates each. For each SNP, the graph depicts the power to detect that SNP at significance level \(\alpha = 5 \times 10^{-8}\). For each SNP, the effect size varies from 0 to 0.05 in increments of 0.001. On the x-axis, we exponentiate effect sizes to covert to odds ratios. See the text for a detailed description of the model