andriusbern / rnagent

RNA design tool that utilizes reinforcement learning

RLIF - RNA Inverse Folding with Reinforcement Learning

This Python package contains:

1. A reinforcement learning based algorithm for RNA inverse folding (82/100 structures solved on the Eterna100 benchmark using various models and folding parameter sets).
2. A graphical interface for RNA design and analysis that integrates RLIF with ViennaRNA package.
3. A command line interface for generating solutions for RNA secondary structures.

GUI of RLIF.

---

Installation:

All the dependencies can be installed by creating a new conda environment from the rlif.yml file and installing the rlif python package:

git clone https://github.com/andriusbern/rlif
cd rlif
conda env create -f rlif.yml
conda activate rlif
pip install -e .

Requirements:

• Conda is required to install this package with the following dependencies:

  1. ViennaRNA==2.4.14
  2. Python=3.6
  3. mpi4py

• The following Python packages are required:

  1. numpy==1.17.1
  2. tensorflow==1.13.1
  3. stable-baselines==2.7.0
  4. pyyaml
  5. gym
  6. forgi
  7. tqdm
  8. PySide2
  9. pyqtgraph
  10. matplotlib

Usage

Graphical User Interface

To launch a Qt based interface:

python rlif/GUI.py

Target secondary RNA structures can be entered using the following methods:

1. Sequence editing field (either nucleotide sequences or secondary structures in dot-bracket notation).
2. Loading a FASTA file containing RNA sequences. Their secondary structures will be predicted using ViennaRNA and can then be used as targets for the algorithm.
3. Loaded from a benchmark dataset (using the dataset selection and Load dataset button).

Command line

To launch a command line interface:

python rlif/CLI.py

Graphical representation of the state and action spaces of the RL model