Manuscript_Avsec_Bioinformatics_2017

Code for Avsec et al, Bioinformatics 2017 (Bioarxiv preprint Avsec et al, Bioarxiv 2017).

Installing requirements

Python

Note, use python>3.5 from anaconda and use virtual environments to not interfere with your default python environment.

pip install -r python3_requirements.txt

R

Start R using: R --vanilla.

Install CRAN R packages:

install.packages(readLines("r_packages.txt"))

Install Bioconductor packages:

source("https://bioconductor.org/biocLite.R")
sapply(readLines("r_bioc_packages.txt"), biocLite)

General repository organization notes

In each experiment folder training predictive models, the main files are:

• readme.md - contains further instructions
• data.py - contains a data() function returning a tuple of train and test-set arrays
• model.py - contains a model() function returning a Keras model
• train.py - runs model training and hyper-parameter optimization

R should be started from the repository root.

All data are located either in Data (smaller clip data) or in data/ (everything else).

Download the data

To download the rest of the data not contained in the repository, run:

wget https://i12g-gagneurweb.in.tum.de/public/paper/Avsec_Bioinformatics_2017/data.tar.gz
tar xvfz data.tar.gz

data/ includes intermediary results, trained models, as well as model training/test datasets.

Support

Let me know if you have any problems by creating an issue or sending me an email to avsec-at-in.tum.de.