Expand description
ESM3 multi-track tokenization.
| Track | Vocab | Layout |
|---|---|---|
| sequence | 33 | <cls>=0 <pad>=1 <eos>=2 ... <mask>=32 |
| structure | 4101 | VQ-VAE 0..4095 + MASK/EOS/BOS/PAD/CB |
| ss8 | 11 | PAD=0 MASK=1 UNK=2 GHITEBSC=3..10 |
| sasa | 19 | PAD=0 MASK=1 UNK=2 bins=3..18 |
Function/InterPro tokenization requires external data files (future work).
Re-exports§
pub use sasa::tokenize_sasa;pub use sequence::decode_sequence;pub use sequence::tokenize_sequence;pub use ss8::tokenize_ss8;pub use structure::tokenize_structure;