DynamICE (Dynamic IDP Creator with Experimental restraints)

A generative to generate new IDP conformer ensembles biased towards experimental data.

Required packages

• Python (3.7.1)
• numpy
• yaml
• pytorch (1.4.0)
• idpconfgen (https://github.com/Oufan75/IDPConformerGenerator.git/tree/grnn, modified from https://github.com/julie-forman-kay-lab/IDPConformerGenerator.git)
• sidechainnet (included, modified from https://github.com/jonathanking/sidechainnet.git)
• X-EISD (https://github.com/THGLab/X-EISD.git)

Training

Modify parameters in training files based on request.

• train_grnn.py for pretraining generative model.
• generate_rnn.py for generating conformers with trained models.
• train_biased_grnn.py Biasing pretrained models with exp data.
• torsion_splits.py for extracting backbone and sidechain torsion angles for pre-training.