Dock2D Doc(k)umentation
The library is used to build a toy protein dataset and models to solve the tasks for molecular recognition:
- Interaction Pose (IP):
Solve for a specific pose between two proteins known to interact. This is the pose generated from the lowest energy transformation that brings a ligand spatially proximal to a receptor.
- Fact-of-Interaction (FI):
Solve for whether two proteins meaningfully interact, if at all. This is the probability of two proteins interacting using the free energy of possible transformations.
Directory Reference
The library is structured into five directories:
Dock2D.DatasetGeneration Generate a protein pool to create IP and FI datasets.
Dock2D.Models Four models, two BruteForce and two Sampling models for the IP and FI tasks each.
Dock2D.Tests Testing scripts to check dataset, algorithm, and model functionality.
Dock2D.Utility Utility classes ranging from data loading to plotting.
Contents:
Cite Us
This code is part of the work done in our paper . Please cite us if you use this code in your own work:
@article{Dock2D,
title={{Dock2D: Synthetic datasets for the molecular recognition problem}},
author={Bhadra-Lobo, Derevyanko, Lamoureux},
journal={arXiv preprint arxiv:2212.03456},
year={2022},
}