Lei and Kai’s work on ProtRAP-LM has been published in Genom. Proteom. Bioinform. (link). The study introduces ProtRAP-LM, a transformer-based model that uses protein language model embeddings to rapidly and accurately predict membrane contact probability (MCP) and relative accessibility for each residue within a given protein sequence. ProtRAP-LM achieves a speed-up of over 300 times compared to MSA-based MCP predictor, enabling proteome-wide predictions within hours. This facilitates large-scale analysis of challenging membrane protein classes, including single-pass transmembrane proteins, membrane-anchored proteins, and β-sheet-containing membrane proteins. In particular, this study provides a comprehensive list of membrane proteins across 48 proteomes, including 78 potential human membrane proteins, offering a valuable resource for future structural and functional studies. An online server is available at http://www.songlab.cn/ProtRAP-LM/home/. Congratulations!