Code for Avsec et al, Bioinformatics 2017 (Bioarxiv preprint Avsec et al, Bioarxiv 2017).
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
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)
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).
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.
Let me know if you have any problems by creating an issue or sending me an email to avsec-at-in.tum.de.