This repository has been created to present the recent work done on drug repositioning thanks to Non-Negative Matrix Factorization [1,2].
The jupyter notebook results.ipynb presents these results.
This repository contains all data, scripts and results related to our recent work. In particular, you will find:
- 2 folders, data which stores the initial data and tmp which stores the results;
- 3 .py files, load_data_NMTF.py, method_NMTF_DatasetContribution.py and method_NMTF.py which create classes used in other files. In particular, the last file contains all methods related to the NMTF;
- 4 other .py files, DatasetContribution.py, dispersion4.py, improvements.py and initialization.py which compute the results presented in the jupyter notebook.
If you want to run these files, you may need to install the following packages:
sklearn, matplotlib, tqdm, scipy, numpy, pandas, seaborn, csv, cs, spherecluster
[1] Dissez, G. and Ceddia G., Pinoli, P. and Ceri, S. and Masseroli, M. (2019). Drug repositioning predictions by non-negative matrix tri-factorization of integrated association data. Proceedings of the 10th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, 25-33.
[2] Ceddia, G. and Pinoli, P. and Ceri, S. and Masseroli, M. (2020). Matrix Factorization-based Technique for Drug Repurposing Predictions. IEEE Journal of Biomedical and Health Informatics, 24(11), 3162-3172.