Additional file 1 of Multi-INTACT: integrative analysis of the genome, transcriptome, and proteome identifies causal mechanisms of complex traits

Additional file 1: Figure S1: Various causal scenarios in which only one type of molecular QTL is colocalized with a causal GWAS hit. Figure S2: EM algorithm estimate distributions across simulated data sets. Figure S3-S5: Distributions of posteriors for each gene product-to-trait effect scenario and three posterior types. Figure S6: Multi-INTACT gene implication results, by tissue, with proportions implicated by marginal analyses. Figure S7: Overlap between genes implicated by TWAS in whole blood and PWAS. Figure S8: Multi-INTACT power stratified by genetic PVEs. Figure S9: Comparison of two multivariate regression test statistic approximations based on summary level-data to the statistic computed from individual-level data. Figure S10: Distributions of expression heritability, protein heritability, complex trait heritability per gene, expression-mediated complex trait heritability per gene, and protein-mediated complex trait heritability per gene in 100 simulated data sets. Figure S11-S13: Multi-INTACT power and realized FDR for simulations. Figure S14: Number of genes tested by Multi-INTACT across tissues. Table S1: Causal diagrams representing possible causal relationships between genotypes, molecular traits and, and a complex trait. Table S2: Simulation INTACT results across a variety of expression and protein prediction models. Table S3: Simulation Multi-INTACT results across a variety of expression and protein prediction models. Table S4: Simulation Multi-INTACT results using both individual-level and summary-level data. Table S5: Multi-INTACT results by varying causal vs. non-causal gene ratios in subsampled simulation data. Supplemental Methods [78–83].