UCSF

Relating proteins based on the similarity of the ligands that bind to and modulate through them is a field that has its basis in classical pharmacology when ligands were tested for phenotypic effects on whole tissues, organs or organisms. We take a ligand-focused view of biology to build protein-protein relationships by describing proteins not by their sequence, structure or function, but by their ligand sets. A systematic approach to relate proteins together both within a family and between families are described. The engine we use to drive a pharmacological organization of proteins is the Similarity Ensemble Approach (SEA) using ligands from ChEMBL to create sets for each protein target as input. We use the output from SEA as a similarity metric to relate proteins pharmacologically.

Family A G protein-coupled receptors (GPCRs) was a family of proteins we first aimed to reorganize pharmacologically. Compared to a sequence based organization of GPCRs, the ligand based organization had a much different arborization of its dendrogram with some GPCRs moving away from their "commonly" related GPCRs while some "distantly" related GPCRs were brought together because they shared common ligands. The pharmacological organization led us to predict for testing GPCRs that were similar based on pharmacology but different from a sequence view. We confirmed three new pairs of GPCRs that were now linked by a new, shared ligand where they previously had no known shared ligands.

Moving beyond GPCRs, we sought a deeper biological relationship between proteins and protein families we related pharmacologically by finding new protein links that not only could potentially share a ligand, based on our predictions, but also shared a phenotype, function, or are implicated in similar diseases. The proteins we related in this manner were also not from the same sequence or structural family but different from classical bioinformatics metrics. The assertion is that because there is a limited number of endogenous signaling molecules, their evolution is almost frozen and therefore, cells and their proteins must evolve and adapt around signaling molecules on top of functioning through them. Relating proteins pharmacologically and phenotypically is explored.