Congratulations to Kai and Lei for the publication of the ProtRAP paper in JCIM (Link). In this paper, we further utilized the membrane contact probability (MCP) from the MemProtMD database and developed a deep learning based predictior that can tell which residues of a given protein sequence are lipid accessible, solvent (water) accessible, or buried in the protein interior. This tool may find wide applications in protein structure prediction and drug discovery, especially when membrane proteins are involved. For quick predictions, an online server has been setup: http://www.songlab.cn/ProtRAP/home/