Internal Coordinate Net (ICoN) for Sampling Conformational Ensembles of Highly Dynamic Proteins via Generative Deep Learning

1. ICoN is trained on ~10000 of fully atomistic and highly flexible conformations

2. It uses vector internal coordinate representation as input features- vBAT

3. It can train ~10K conformations in a few mins, and generate ~100K conformations in less than a min.

Dependancies

• python 3.8 >
  • pytorch - for Deep Learning
  • MDAnalysis - for trajectory I/O
  • pytraj - for analysis

Installation

• Clone the repository: git clone https://github.com/chang-group/ICoN.git
• Navigate to corresponding folder for further instruction