This is the work of team CSU_CHINA-iDEC:
- The one model for peptide genesis
This program is also a team-competition work in International Directed Evolution Competition(abbr. into iDEC)
Peplib Generator: In silico directed evolution and de novo generation for
peptide drug powered by AlphaFold2 || Supplementary Data
Other work-related paper will be listed later
Due to job adjustment, our work has been modified and integrated into a new pipeline for high-throughput in silico generation and screening:
- Rif sampling: by RifGen/our sampling scripts(unreleased yet)
- Scaffold generation: WeFold(unreleased yet)
- Scaffold filtering: Molpacker(unreleased yet)
- Sequence design: FastDesign/ProteinMPNN special edition edited from ProteinMPNN
- Sequence Filtering: AlphaFold2-turbo(from Peplib Generator edited from AlphaFold2, unreleased yet)
It takes 10-15 days to generate ideal 20k-100k sequences & structures.
- Hardware(supported SLURM)
- CPU: >=1000 * x86-64 cores; arm64 based is not supported
- GPU: >=10 * NVIDIA® Tesla® P100 or higher, with 16GB+ VRAM
- RAM: 15 GB each GPU core, 4GB each CPU core
- ROM: 500GB or higher