Implementation of Bipartite Graph-represented Drug Response Predictor (BiG-DRP and BiG-DRP+) as described in:
David Earl Hostallero, Yihui Li, Amin Emad, Looking at the BiG picture: incorporating bipartite graphs in drug response prediction, Bioinformatics, Volume 38, Issue 14, 15 July 2022, Pages 3609โ3620, https://doi.org/10.1093/bioinformatics/btac383
This repository has been tested on python 3.7. To install the dependencies run the following on the terminal
pip install -r requirements.txt
python main.py
To run BiG-DRP+, you must first run BiG-DRP while specifying the results subfolder (--folder=<folder_name>
). Then run BiG-DRP with the --weight_folder
specified as the results subfolder in the previous run.
python main.py --mode=train --folder=big
python main.py --mode=extra --weight_folder=big --folder=big_plus
--split
: the type of data-splitting to use (lco
orlpo
, default:lco
)--dataroot
: the root directory of your data (file names for input files can me modified inutils/constants.py
) (default:../
)--outroot
: the root directory of your outputs (default:./
)--folder
: subdirectory you want to save your outputs (optional)--weight_folder
: subdirectory for the saved weights and encodings (for BiG-DRP+ only)--mode
:train
means BiG-DRP,extra
means BiG-DRP+ (default:train
)--seed
: the seed number for 5-fold CV (default: 0)--drug_feat
: type of drug feature (desc
,morgan
, ormixed
, default:desc
)--network_perc
: percentile used for the bipartite graph threshold (default: 1)
Preprocessed data can be accessed here: https://dx.doi.org/10.6084/m9.figshare.20022947
@article{hostallero2022looking,
title={Looking at the BiG picture: incorporating bipartite graphs in drug response prediction},
author={Hostallero, David Earl and Li, Yihui and Emad, Amin},
journal={Bioinformatics},
volume={38},
number={14},
pages={3609--3620},
year={2022},
publisher={Oxford University Press}
}