Abstract

Background: The PHOME (PHarmacology + proteOME) is a drug re-purposing algorithm that exploits large publicly available datasets to identify novel drug/target interactions in platelets. The PHOME integrates pharmacological, proteomic, tissue expression and gene function data into interaction networks.
Aims: (1) To use the PHOME to rank known compounds according to their ability to modulate platelet function. (2) To use in vitro platelet testing to evaluate the predictive effectiveness of the PHOME in identifying compounds with platelet modulatory actions.
Methods: 1. Drug Selection using the PHOME: A platelet protein-protein interaction (PPI) network was constructed using the Gene Expression Atlas (https://www.ebi.ac.uk/gxa/home) and STRING (https://string-db.org/) databases, with drug/target annotations from STITCH (http://stitch.embl.de/). We scored drug effects in the Platelet-PPI-Network generating outputs for 553 well-known and/or clinically used compounds. Lower P values (Wilcoxon rank-sum test) predict effects on collagen binding (GO:0005518), platelet activation (GO:0030168) and platelet aggregation (GO:0070527). Ten compounds scored P=1 in each GO category and were selected for testing as controls. Ten compounds were selected randomly from 37 that scored P=0 in each category.
2. Drug Evaluation: These 20 drugs were assigned ID codes to remove operator bias. Human washed platelets were pre-incubated with the 20 drugs at 100 µM or 4 vehicle controls. Collagen (1 µg/ml)-induced aggregation and static adhesion to collagen, collagen-related peptide (CRP) and fibrinogen were measured.
Results: 5/10 P=1 drugs substantially inhibited aggregation and 5/10 had no discernible effect. 6/10 P=0 drugs inhibited aggregation and 4/10 did not. 3/10 P=1 drugs reduced adhesion to collagen, 1/10 to CRP and 2/10 to fibrinogen. For the P=0 drugs, the equivalent results were 1/10, 0/10 and 1/10.
Conclusion(s): These data suggest that the PHOME does not effectively predict drugs that modify platelet function, although 11 compounds displayed inhibitory actions. Scrutiny of database entries may suggest novel hypotheses about platelet function and enable algorithm optimisation.