Identification and functional analysis of a serine protease inhibitor using machine learning strategy

Abstract

In the intricate realm of animal biology, a multitude of vital processes heavily rely on precisely orchestrated proteinase cascades, but the potential for havoc makes proteinase inhibitors indispensable, with serine proteinase inhibitors (serpins) at the forefront, serving as custodians of homeostasis and participating in various critical biological processes. Importantly, there are still many unexplored facets of serpin functionality. In this study, we focused on the serpin family proteins from Marsupenaeus japonicus, utilizing a fine-tuned pretrained protein language model. This approach led to the identification and evolutionary validation of 28 serpins, one of which, referred to as Mj_serpin-1, was both computationally and experimentally demonstrated to show potential as an antiviral and apoptosis inhibitor. Our research unveils exciting prospects for the fusion of state-of-the-art artificial intelligence and rich bioinformatics, holding the promise of significant discoveries that could pave the way for future therapeutic advancements.

Keywords: Antivirals; Apoptosis inhibitor; Large protein language model; Molecular docking; Serine proteinase inhibitor.