ProtRAP-LM: Fast and accurate protein relative accessibility prediction and membrane protein screening through protein language model embeddings

Abstract Membrane proteins play pivotal roles in cellular signaling and transport, making them prime targets for therapeutic intervention. Therefore, expeditious screening and accurate property prediction of these proteins is crucial. Recently, we proposed a new metric, the membrane contact probability (MCP), to characterize the membrane-contacting features of membrane proteins and further refine the prediction of the relative accessibility of proteins (ProtRAP). However, these models relied on evolutionary information in the form of multiple sequence alignments (MSAs), which hindered rapid predictions. In this study, we present a novel transformer-based model, ProtRAP-LM, utilizing language model (LM) embeddings as input features, to quickly and accurately predict MCP and relative accessibility for each residue of a given protein sequence. ProtRAP-LM can achieve accurate predictions for entire proteomes within hours, demonstrating superior performance compared to previous MSA-based models, with a speedup of over 300 times. This empowers us to furnish more thorough annotations of membrane protein sequences on a proteome-wide scale, particularly for single-pass transmembrane proteins, membrane-anchored proteins, and β-sheet-containing membrane proteins, which have long been a challenge in the field. In the end, we provide a comprehensive list of membrane proteins for 48 living organisms, offering a rich resource for investigating the structure and function of these essential biomolecules in the future. Competing Interest Statement The authors have declared no competing interest.