Omic Risk Scores are Associated with COPD-related Traits Across Three Cohorts

Figures

• 

  Figure 1.. Performance of ORS in the COPDGene testing cohort. A) Score significance plotted against effect size for ORS associations with traits collected concurrently. Traits modeled with a Gaussian distribution are plotted with standardized β coefficient as the effect size measure while non-Gaussian-model traits are plotted with odds ratio. B) Adjusted R2 of trait models with and without ORS for traits measured concurrently with omics. Grey bars indicate variance explained by models including only clinical risk factors. AUROC of trait models with and without ORS.

• 

  Figure 2.. ORS improve prediction of COPD status. A) Violin plots of the FEV1/FVC MetRS, ProtRS, and TransRS stratified by COPD cases and controls. B) ROC plot of COPD predictive models including clinical covariates with or without the MetRS, ProtRS, or TransRS. C) Top 20 features in the MetRS, ProtRS, and TransRS, ranked by feature importance. Bars are colored by direction of association.

• 

  Figure 3.. Replication of ORS in SPIROMICS & MESA. A) Score significance plotted against effect size for ORS tested in SPIROMICS. B) Standardized β coefficients and 95% confidence intervals for Gaussian trait model ORS in SPIROMICS. C) Odds ratios for binomial trait model ORS in SPIROMICS. D) Adjusted R2 for Gaussian trait model ORS in SPIROMICS. E) AUROC for binomial trait model ORS in SPIROMICS. F) Score significance plotted against effect size for ORS tested in MESA. G) Standardized β coefficients and 95% confidence intervals for Gaussian trait model ORS in MESA. H) Odds ratios for binomial trait ORS in MESA. I) Adjusted R2 for Gaussian trait model ORS in MESA. J) AUROC for binomial trait model ORS in MESA.

• 

  Figure 4.. ORS of Progression Traits and Mortality in COPDGene. A) Standardized β coefficients and 95% confidence intervals for ORS. Darker bars indicate significant associations. B) Adjusted R2 for significant ORS in addition to clinical risk factors. C) Kaplan-Maier survival curve colored by ProtRS quartiles. D) Kaplan-Maier survival curve colored by MetRS quartiles. E) Kaplan-Maier survival curve colored by TransRS quartiles.

• 

  Figure 5.. MultRS of Continuous COPDGene Traits. A) Standardized β coefficients and 95% confidence intervals for ORS and MultRS. B) Adjusted R2 for MultRS and ORS in addition to clinical risk factors. C) Top 60 features in the lung density MultRS. Positive associations are darkly shaded. D) Top 20 features in the lung density MetRS. E) Top 20 features in the lung density ProtRS. F) Top 20 features in the lung density TransRS.