Svsbi: Sequence-based Virtual Screening Of Biomolecular Interactions

This article proposes a sequence-based visual screening (SVS) of biomolecular interactions that can predict a wide variety of biological interactions at structure-level accuracy without invoking 3D structures. The SVS framework consists of multiple NLP models, extracts evolutionary, and contextual information from different biomolecules simultaneously to reconstruct sequence representations for interactive molecules, such as proteins, nucleic acids, and/or small molecules. Extensive validations indicate that SVS is a general, accurate, robust, and efficient method for the virtual screening of biomolecular interactions.