Flowchart of the 3 stages of the analysis.

Stage 1: Separate analyses of full RNAseq and proteomics datasets (Liver 3-Way RNAseq Full, Liver 3-Way Proteomics Full, PBMC 3-Way RNAseq Full, and PBMC 3-Way Proteomics Full). To simplify the flowchart, we are only showing one representative dataset, which we will refer to as “3-Way Full Datasets”. Stage 2: Training ML models in unmatched balanced data with subsequent testing and integration in matched balanced data. Part A: Identification of top transcriptomic and proteomic pipeline configurations along with their corresponding gene and protein sets for unmatched balanced datasets. Part B: Evaluation of top performing models with their corresponding gene and protein sets from part A in matched balanced data. Part C: Integration of paired sets of the top performing gene and proteomics models with their corresponding gene and protein sets, in matched balanced data. Stage 3: Intersection analysis of the combined best gene-protein sets for liver samples and for PBMCs.