https://github.com/peldom/peplib_generator

The one model for genesis of peptide ligands
https://github.com/peldom/peplib_generator

aidd cadd deep-learning peptide-prediction text-generator

Last synced: 6 months ago
JSON representation

The one model for genesis of peptide ligands

• Host: GitHub
• URL: https://github.com/peldom/peplib_generator
• Owner: Peldom
• License: mit
• Created: 2021-05-27T04:33:13.000Z (over 4 years ago)
• Default Branch: main
• Last Pushed: 2022-10-24T00:44:23.000Z (about 3 years ago)
• Last Synced: 2025-01-05T20:42:17.775Z (about 1 year ago)
• Topics: aidd, cadd, deep-learning, peptide-prediction, text-generator
• Homepage:
• Size: 14.8 MB
• Stars: 8
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Peplib Generator

![Team Poster](https://github.com/Peldom/Peplib_Generator/blob/main/README_Support/Poster_CSU_CHINA_v2.png)

--------------------------------------------------------------------------------

This is the work of team CSU_CHINA-iDEC:

![Team Logo](https://github.com/Peldom/Peplib_Generator/blob/main/README_Support/Teamlogo.png)

- The one model for peptide genesis

  

## [`Team wiki`](https://idec2021.github.io/CSU_CHINA/idec/home.html)

## [`iDEC wiki`](http://idec.io)

This program is also a team-competition work in ***International Directed Evolution Competition***(abbr. into ***iDEC***)

## Pre-Print

[Peplib Generator: In silico directed evolution and de novo generation for

peptide drug powered by AlphaFold2](https://arxiv.idec.io/article/000004/) || [Supplementary Data](https://arxiv.idec.io/pdf/a000004.01.pdf)  

Other work-related paper will be listed later

## Workflow

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](https://github.com/LongxingCao/rifdock_v4.2)/our sampling scripts(unreleased yet)

- Scaffold generation: WeFold(unreleased yet)

- Scaffold filtering: Molpacker(unreleased yet)

- Sequence design: [FastDesign](https://www.rosettacommons.org/docs/latest/scripting_documentation/RosettaScripts/Movers/movers_pages/FastDesignMover)/**ProteinMPNN special edition** edited from [ProteinMPNN](https://github.com/dauparas/ProteinMPNN)

- Sequence Filtering: **AlphaFold2-turbo**(from **Peplib Generator** edited from [AlphaFold2](https://github.com/lucidrains/alphafold2), unreleased yet)

It takes 10-15 days to generate ideal 20k-100k sequences & structures.

## Hardware Dependencies

- 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