Back

SARS-CoV-2 PLpro whole human proteome cleavage prediction and enrichment/depletion analysis

Prescott, L.

2021-10-06 bioinformatics
10.1101/2021.10.04.462902 bioRxiv
Show abstract

A novel coronavirus (SARS-CoV-2) has caused a pandemic that has killed millions of people, worldwide vaccination and herd immunity are still far away, and few therapeutics are approved by regulatory agencies for widespread use. The coronavirus 3-chymotrypsin-like protease (3CLpro) is a commonly investigated target in COVID-19, however less work has been directed toward the equally important papain-like protease (PLpro). PLpro is less characterized due to its fewer and more diverse cleavages in coronavirus proteomes and the assumption that it mainly modulates host pathways with its deubiquitinating activity. Here, I extend my previous work on 3CLpro human cleavage prediction and enrichment/depletion analysis to PLpro.[1] Using three sets of neural networks trained on different taxonomic ranks of dataset with a maximum of 463 different putative PLpro cleavages, Matthews correlation coefficients of 0.900, 0.948, and 0.966 were achieved for Coronaviridae, Betacoronavirus, and Sarbecovirus, respectively. I predict that more than 1,000 human proteins may be cleaved by PLpro depending on diversity of the training dataset and that many of these proteins are distinct from those previously predicted to be cleaved by 3CLpro. PLpro cleavages are similarly nonrandomly distributed and result in many annotations shared with 3CLpro cleavages including ubiquitination, poly(A) tail and 5 cap RNA binding proteins, helicases, and endogenous viral proteins. Combining PLpro with 3CLpro cleavage predictions, additional novel enrichment analysis was performed on known substrates of cleaved E3 ubiquitin ligases with results indicating that many pathways including viral RNA sensing are affected indirectly by E3 ligase cleavage independent of traditional PLpro deubiquitinating activity. As with 3CLpro, PLpro whole proteome cleavage prediction revealed many novel potential therapeutic targets against coronaviruses, although experimental verification is similarly required.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.