This repository is based on the ColabFold on GitHub.
Green shows the predicted structure of the PDB target protein, cyan shows the predicted structure of target protein by using AlphaFold-Multimer model, and magenta shows the predicted structure of cyclic peptide by using AlphaFold-Multimer model.
The same input format can be used as when predicting complexes with ColabFold. However, you can predict protein-cyclic peptide complexes by using the target protein sequence as input before ":" and the cyclic peptide sequence after ":" as in TARGETPROTEINSEQ:CYCLICPEPTIDESEQ and check the set_cyclic option.
AlphaFold training dataset has not been processed to remove cyclic peptides, but there is a cutoff based on the number of amino acid residues. In the latest version of the alphafold2_multimer_v3 model training set, peptides less than 4 amino acid residues were excluded. However, the alphafold2_ptm model and the alphafold2_multimer_v2 model training set, peptides less than 16 amino acid residues were excluded (confirmed with the AlphaFold Team at DeepMind).
| moldels | Length of residues | Released date for PDB structures |
|---|---|---|
| alphafold2_ptm | 16 or more residues | before 2018-04-30 |
| alphafold2_multimer_v1 | 16 or more residues | before 2018-04-30 |
| alphafold2_multimer_v2 | 16 or more residues | before 2018-04-30 |
| alphafold2_multimer_v3 | 4 or more residues | before 2021-09-30 |
Kosugi, T.; Ohue, M. Design of Cyclic Peptides Targeting Protein–Protein Interactions Using AlphaFold Int. J. Mol. Sci. 2023, 24, 13257. doi:10.3390/ijms241713257
Kosugi, T.; Ohue, M. Design of Cyclic Peptides Targeting Protein-Protein Interactions Using AlphaFold bioRxiv 2023, doi:10.1101/2023.08.20.554056.