TMVisDB - Database Release This is the database release accompanying the paper "TMVisDB: A Resource for Transmembrane Protein Annotations and 3D Visualizations" (2025). Description TMVisDB is a comprehensive SQLite database containing annotations for over 46 million predicted transmembrane proteins. The database combines AlphaFold DB structure predictions with transmembrane topology predictions from the protein language model (pLM) based method TMbed. This release contains the core database structure and annotations, with sequence information and database indices removed due to size constraints. The full database with sequences and indices is available through the web interface at https://tmvisdb.rostlab.org. Database Contents: TMbed prediction Taxonomy (Domain) Transmembrane type Signal Peptide All Bacteria Eukaryota Archaea Other Alpha-helix (TMH) no 39,391,449 28,020,596 9,944,469 991,477 434,907 yes 4,475,707 2,977,947 1,306,796 152,801 38,163 Beta-stranded barrels (TMB) no 323,017 269,126 43,215 1,401 9,275 yes 1,814,523 1,786,881 7,959 1,608 18,075 Both no 20,918 16,937 3,584 90 307 yes 22,836 21,731 771 52 282 Total 46,048,450 33,093,218 11,306,794 1,147,429 501,009 Schema The database consists of the following main tables: organism: Contains taxonomic information sequence: Links proteins to their organisms and UniProt identifiers annotation: Stores transmembrane topology predictions tminfo: Contains transmembrane statistics and classifications A diagram of the database scheme can be found here Note on Removed Data To make the database distribution feasible, we removed: All sequence information from the sequence table All database indices (which can be recreated using the schema) Users requiring sequence information should: Reference the UniProt accessions provided in the database Access the full database through our web interface Or rebuild indices locally if needed using the schema provided in our GitHub repository Associated Resources Web Interface Source Code Database Scheme