Personalized Drug Target Inference
Taking a personalized approach shows potential improvement over classical inference methods.
A standard approach to identify proteins as potential drug target candidates is to perform gene knockouts on model organisms, essential disabling the gene and examining the effect on the organism and the disease.
By applying cancer patient-specific genetic mutation and expression data on protein-protein interaction networks, and performing computer simulations of knockouts - we are able to predict known drug targets and propose novel ones. This workflow and approach could potentially allow more personalized, effective treatment.
This work was done as part of the Computational Biology workshop at Tel-Aviv University and supervised by Prof. Roded Sharan. Idea, implementation, and evaluation was done in equal parts by my partner, Ortal Shnaps, and myself.
Published in Pacific Symposium of Biocomputing 2016 Proceedings.
Code available here.
Explaining the work to innocent bystanders at PSB 2016, Big Island of Hawaii