Improvement: Integrate proteome-transcriptome ratios (human tissues)

[dagl1]

Your method utilizes RNA expression, which doesn't necessarily relate 1 to 1 with protein levels. An existing set of per-gene per-tissue ratio of mRNA to protein exists (and it is trivial to multiply your omics data with this PTR dataset before applying your GPR rules). This might lead to even better predictions and could also potentially help with the few metabolites which appeared to not align as well. The observed ratios span 10e2 to 10e6, and are relatively stable across tissues, but quite different across genes, thus high PTR genes might only increase by small amounts, but actually have a disproportionate affect.

Study: https://www.embopress.org/doi/full/10.15252/msb.20188513 dataset is found in supplemenatry table EV3 (contains mRNA, protein, and PTR all in one sheet)