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

We use the information available from other viruses that have infected humans (e.g. 2003 SARS virus, Dengue virus, Zika virus) to predict how physical interactions between human proteins and those from the novel coronavirus. We leveraged the Graham supercomputer to predict every possible relationship (>280,000) between nCoV and humans and generated these predictions. The algorithms are also used to identify the specific parts of the viral proteins that likely cause these interactions which is potentially useful to design drugs that can block that mechanism. This computational work is meant to function as a comprehensive "guide" to support other researchers in their exploration of new ways to protect against the novel coronavirus. Figure 10 : Compilation of the One-to-All Score Curves for each SARS-CoV-2 protein by PIPE4 (blue) and SPRINT (green). Each of the subplots depicts a characteristic "L"-shape, where there are a relatively small number of high-scoring pairs as compared to a large number of low-scoring pairs within the baseline. Note that the y-axes are not shared among subplots.