Computational Prediction of the Comprehensive SARS-CoV-2 vs. Human Interactome to Guide the Design of Therapeutics

Kevin Dick; Kyle K Biggar; James R Green

Thus, for each one-to-all score curve, a score threshold delineating the "high-scoring" PPIs from the baseline was identified and used to determine the high-confidence interactions. In the absence of known PPIs between SARS-CoV-2 and human, it is difficult to determine a suitable global decision threshold. By instead examining the morphology of the one-to-all score curves for both perspectives, we can qualitatively identify high-scoring pairs. This process can be further automated through the identification of the baseline/knee for each view under the assumption that true PPIs are rare and high-scoring, while non-interacting pairs tend to generate scores residing below the knee in the baseline. In Figure 2 , we overlay the one-to-all score curves for each SARS-CoV-2 protein and "zoom' into the high-score/low-rank region to emphasize that the selection of a single global top-k or score threshold would inappropriately exclude relatively high-scoring pairs within specific SARS-CoV-2 proteins, while admitting too many low-scoring putative PPI for other proteins.