We enable the community to query and visualise our results interactively through our webserver: omicscience.org/apps/protdet…

Our data can enhance interpretation of the ever-increasing number protein-disease associations. E.g. proteins associated with incident #PAD are enriched in those casually determined by #smoking status.

We performed #Mendelian randomization to identify major biological influences with putative causal effects on plasma protein levels.

#Proteins influenced by one or few strong factors can serve as biomarkers of health-related changes and exposures, e.g. #SOST for bone mass, #CTHRC1 for #lipid-lowering medications.

Liver and kidney function, inflammation and smoking status have wide-spread effects on the proteome (influencing >3k proteins). Anthropometric factors influenced a fewer number of proteins but had a stronger influence.

Joint genetic and biological factors explained >10% of variation for over a third of proteins. #Plasma proteins cluster into groups according to their major biological explanatory factor. Over a third of proteins were best explained by #modifiable factors.

#Proteomics capture thousands of plasma proteins. Most are not secreted, their origin and role are unclear. We characterised the joint contribution of >60 biological, #genetic and technical influences on 5k #plasma proteins in the Fenland study.

We make our results available through our webserver (omicscience.org/apps/protpre…) to visualise the potential benefit of #proteomic screening in a Bayesian framework.

Severe diseases with poor prognosis among the best predicted: dilated cardiomyopathy, pulmonary fibrosis, #MultipleMyeloma , non-hodgkin lymphoma.

We find proteins predictive across more than 1 clinical specialty and specific disease predictors e.g. BCMA (TNFRSF17) for multiple myeloma, a target of multiple myeloma treatments.