INSTALLATION

• Clone the repository
• Run pip install . in the root directory. It's recommended to create a virtual environment first, use rnadeep.yaml for that (e.g. conda env create -n rnadeep --file rnadeep.yaml).

DIRECTORIES

• rnadeep: The nessecary tools to sample and encode the learning data, the spotrna models, and important metrics.

• rnaconv: Contains the rfam database as well as multiple tools to generate artifical data in different ways using RNAfold³ and SISSI⁴. Refer to rnaconv/ReadMe.md

• examples: Example python, bash and slurm scripts to train and predict data. Refer to examples/ReadMe.md

• _{³https://github.com/ViennaRNA}
  

• _{⁴https://cibiv.github.io/software/sissi}
  

DOCUMENTATION

• For a documentation document of the modules and functions refer to doc/documentation.pdf

Original Literature:

Caveats to deep learning approaches to RNA secondary structure prediction

Christoph Flamm¹, Julia Wielach¹, Michael T. Wolfinger^1,2, Stefan Badelt¹, Ronny Lorenz¹, Ivo L. Hofacker^1,2

_{¹Department of Theoretical Chemistry, University of Vienna, Vienna, Austria
²Research Group Bioinformatics and Computational Biology, Faculty of Computer Science, University of Vienna, Vienna, Austria}

This repository contains additional resources that were used during preparation of the manuscript. These include custom code, Google Colab notebook and data files.