We finally bring it all together in a knowledge graph, to demonstrate which proteins are good indicators of drug treatment.

Explore here omicscience.org/apps/prot_fo...

We bring this resources directly to your study via enrichment analysis, demonstrating hidden insights and residual confounding among published studies.

github.com/comp-med/r-p...

We demonstrate how this knowledge can guide biomarker identification, reducing candidate lists by almost 40-fold.

This included disease-specific findings like a >2.5-fold increased risk for abdominal aortic aneurysm associated with plasma levels of MMP12.

Not all people are the same, any many factors explaining variance differ across the sexes, but also genetically-inferred ancestry.

Some genetic variants much more important in people of Non-European ancestry.

We demonstrate that some of the variation can be traced back to specific tissues and cell-types, including #Fibroblasts as so far under-appreciated contributing cell-type.

We identify different axes along proteins vary, incorporating >1.7k variables.

We explain >25% of variation for >50% of proteins.

Modifiable factors explained >> non-modifiable factors -> proteins are great as dynamic read out