APORC Document Center: DrugRepos

More...

PreDR: Drug repositioning from chemical, genomic and pharmacological data in an integrated framework

Version 0.0.1
Last updated: Oct. 16, 2012

References#

Yong-Cui Wang, Nai-Yang Deng, and Yong Wang. Drug repositioning from chemical, genomic and

pharmacological data in an integrated framework. 'Bioinformatics, in revision.

Motivation#

Computational inference of new diseases for existing drugs (drug repositioning) offers the possibility of faster drug development and can also reduces risks in drug discovery. Previous research indicated that drugs with similar chemical structures, target proteins, or side-effects may always treat similar diseases. However, each single data source is important in each own way and data integration is crucial to reposition drug more accurately.

Method#

Here, we propose a novel method named PreDR, to integrate chemical structures, target proteins, and side-effects to predict drug repositioning by uncovering the potential associations between drugs and diseases. Specially, we first characterize drug and disease by their similarity-based profiles, and define the kernel function to correlate drug with disease. Finally, we train a machine learning model, support vector machine (SVM), to automatically predict novel drug-disease interactions. The new method is validated on a well-established drug-disease networks with 1,933 interactions between 593 drugs and 313 disease. By cross-validation, we find that all chemical structures, drug targets, and side-effects information are predictive for drug-disease interactions prediction. Moreover, by integrating these three properties, more experimentally observed drug-disease interactions can be revealed. In addition, database research and disease gene pathway analysis indicate that our new predictions are worthy of future experimental validation. We further test our novel prediction under clinical trials, and show the significant prospect of our new method in future drug treatment.