KnotFold

About The Project

The implementation of the paper "Accurate prediction of RNA secondary structure including pseudoknots through solving minimum-cost flow with learned potentials".

Getting Started

Prerequisites

Install PyTorch 1.6+, python 3.7+

Installation

1. Clone the repo

   git clone https://github.com/gongtiansu/KnotFold.git

2. Install python packages

   cd KnotFold
   pip install -r requirements.txt

Usage

1. KnotFold.py: predicting RNA secondary structure (bpseq format) from RNA sequence (fasta format).

   python KnotFold.py -i <RNA_fasta> -o <output_dictionary> (--cuda)

2. KnotFold_mincostflow: constructing RNA secondary structure from base pairing probability using the minimum-cost flow algorithm.

   KnotFold_mincostflow <prior_probability> <reference_probability> 

3. KnotFold_mincostflow.cc: the source C++ code of the minimum-cost flow algorithm.

   g++ KnotFold_mincostflow.cc -o KnotFold_mincostflow -std=c++0x -O2 

   Example

   cd example
   ./run_example.sh

   Training

   We provide training code for KnotFold.

License

Distributed under the MIT License. See LICENSE for more information.