DVA

"A computational approach to aid clinicians in selecting anti-viral drugs for COVID-19 trials"

Aanchal Mongia (IIIT Delhi), Sanjay Kr. Saha (IPGMER, Calcutta), Emilie Chouzenoux (OPIS, Inria Saclay), Angshul Majumdar (IIIT Delhi)

User-friendly prediction tool available at: http://dva.salsa.iiitd.edu.in/

This repository (DVA) contains:

• DVA (Drug virus association database)
• Collection of matrix completion based computational techniques to predict anti-viral drug prediction for viruses

Sources

• DrugBank: https://www.drugbank.ca/categories/DBCAT000066
• Antiviral drugs for viruses other than human immunodeficiency virus
• Approved antiviral drugs over the past 50 years
• Long-acting neuraminidase inhibitor laninamivir octanoate (cs-8958) versus oseltamivir as treatment for children with infuenza virus infection.
• Effectiveness of chloroquine and inflammatory cytokine response in patients with early persistent musculoskeletal pain and arthritis following chikungunya virus infection
• Heat shock protein 90 positively regulates chikungunya virus replication by stabilizing viral non-structural protein nsp2 during infection.
• Chikungunya virus: in vitro response to combination therapy with ribavirin and interferon alfa 2a.
• Structural basis for the inhibition of covid-19 virus main protease by carmofur, an antineoplastic drug
• Repurposing of the anti-malaria drug chloroquine for zika virus treatment and prophylaxis.
• Potential benefts of ibuprofen in the treatment of viral respiratory infections.
• ViPR: http://www.viprbrc.org/

The raw data can be found at: ./data_raw/database.xlsx. The processed data has been created using the notebook read_database.ipynb. A schematic view of the DVA database curation and association prediction using it has been shown below.

The computational algorithms used to predict drug-virus association are available in: helper_functions/alg_template. These are:

• Nuclear Norm Minimization based matrix completion [1]
• Matrix Facrorization based matrix completion [1]
• Deep matrix factorization [2]
• Graph regularized matrix factorization [3]
• Graph regularized matrix completion [4]
• Graph regularized binary matrix completion [5]

The results in the paper above can be reproduced by the following MATLAB scripts:

• run.m
• ./Experiments/novel_drugs_prediction.m
• ./Experiments/coronavirus_pred.m

References

[1] Mongia, Aanchal, Debarka Sengupta, and Angshul Majumdar. "McImpute: Matrix completion based imputation for single cell RNA-seq data." Frontiers in genetics 10 (2019): 9.

[2] Mongia, Aanchal, Debarka Sengupta, and Angshul Majumdar. "deepMc: Deep Matrix Completion for Imputation of Single-Cell RNA-seq Data." Journal of Computational Biology (2019).

[3] Ezzat, Ali, et al. "Drug-target interaction prediction with graph regularized matrix factorization." IEEE/ACM transactions on computational biology and bioinformatics 14.3 (2016): 646-656.

[4] Mongia, Aanchal, and Angshul Majumdar. "Drug-target interaction prediction using Multi Graph Regularized Nuclear Norm Minimization." Plos one 15.1 (2020): e0226484.

[5] Mongia, Aanchal, Emilie Chouzenoux, and Angshul Majumdar. "Computational prediction of Drug-Disease association based on Graph-regularized one bit Matrix completion." bioRxiv (2020).

Cite us:

@article{mongia2020computational,
 title={A computational approach to aid clinicians in selecting anti-viral drugs for COVID-19 trials},
 author={Mongia, Aanchal and Saha, Sanjay Kr and Chouzenoux, Emilie and Majumdar, Angshul},
 journal={arXiv preprint arXiv:2007.01902},
 year={2020}
}