auto-dp

This repository contains proof-of-concept code for the methods described our WABI 2022 paper. It consists of a snakemake workflow, that is capable of producing various types of output files (latex equations, visualizations...) given a user-specified input file.

As described in the paper, the output latex equations are for a DP scheme solving the folding problem restricted to the family of RNA structures specified by an input fatgraph. This input fatgraph needs to be written in a file in a specific directory of the repository, while the outputs are then generated with snakemake terminal commands.

Dependencies

The following are needed to execute the snakemake pipeline:

• snakemake.

• a Java compiler for the meiji tree decomposition solver (source code included in this repo under workflow/scripts/PACE2017-TrackA, see compilation section below).

• a TeX distribution (capable of executing pdflatex commands).

This code was developed on a Linux machine. As snakemake will execute commands in the terminal directly, the commands must be execute on a Unix-like terminal in order to work.

How to use

Imagine you are interested in the following kissing-hairpins pattern: ([)(]).

Such a string will describe a fatgraph in which each base pair is seen as an helix of arbitrary length. It needs to be written in a one-line file called, for instance, khp.dbn and located under resources/dbn_files.

Latex equations such as the ones presented in our paper can then be produced with:

snakemake -c1 results/latex_equations/khp_latex_equations.pdf

Given this command, snakemake will look for rules to produce results/latex_equations/khp_latex_equations.pdf. It will see that it is capable of producing results/latex_equations/{name}_latex_equations.pdf given (with a few steps in between) resources/dbn_files/{name}.dbn.

Meiji solver compilation

cd workflow/scripts/PACE2017-TrackA/
javac tw/exact/*.java