Reference database
conodictor-db-2026_04, built from UniProtKB release 2026_04 by conodb 0.1.0.
The database is rebuilt from UniProtKB/Swiss-Prot with conodb: reviewed Conus conopeptide precursors, curated superfamily labels, entries grouped at 70% identity so that no near-duplicate is shared between training and evaluation, region profiles (HMM, PSSM, position weight matrices), full-precursor models for region prediction, and a multinomial classifier with a NONE class trained against non-conopeptide secreted proteins and shuffled precursors.
Superfamilies (18)
| Superfamily | Mature peptides in the database |
|---|---|
| A | 201 |
| B | 14 |
| C | 18 |
| Con-insulin | 50 |
| D | 33 |
| H | 5 |
| I1 | 38 |
| I2 | 32 |
| I3 | 8 |
| J | 8 |
| L | 8 |
| M | 187 |
| O1 | 289 |
| O2 | 113 |
| O3 | 22 |
| P | 8 |
| S | 9 |
| T | 154 |
Classifier features: SIG:hmm, SIG:pssm, SIG:pwm, PRO:hmm, PRO:pssm, PRO:pwm, MAT:hmm, MAT:pssm, MAT:pwm, FW:framework.
Every release, its evaluation and the tables behind it are archived on Zenodo; the evaluation, benchmark and supplementary tables are listed in the paper.