TMbed: transmembrane proteins predicted through language model embeddings - Precomputed Predictions

We provide precomputed predictions for the human proteome and for UniProtKB/Swiss-Prot.

• Human (21-04-2022)
• UniProtKB/Swiss-Prot (11-05-2022)

Data Format

Regular text file with single letters encoding predicted residue classes.
In 3-line format, every protein is represented by three lines: header, sequence, labels.

• B: Transmembrane beta strand (IN-->OUT orientation)
• b: Transmembrane beta strand (OUT-->IN orientation)
• H: Transmembrane alpha helix (IN-->OUT orientation)
• h: Transmembrane alpha helix (OUT-->IN orientation)
• S: Signal peptide
• .: Non-Transmembrane