Drug repositioning: a machine-learning approach through data integration

Abstract

Existing computational methods for drug repositioning either rely only on the gene expression response of cell lines after treatment, or on drug-to-disease relationships, merging several information levels. However, the noisy nature of the gene expression and the scarcity of genomic data for many diseases are important limitations to such approaches. Here we focused on a drug-centered approach by predicting the therapeutic class of FDA-approved compounds, not considering data concerning the diseases. We propose a novel computational approach to predict drug repositioning based on state-of-the-art machine-learning algorithms. We have integrated multiple layers of information: i) on the distances of the drugs based on how similar are their chemical structures, ii) on how close are their targets within the protein-protein interaction network, and iii) on how correlated are the gene expression patterns after treatment. Our classifier reaches high accuracy levels (78%), allowing us to re-interpret the top misclassifications as re-classifications, after rigorous statistical evaluation. Efficient drug repurposing has the potential to significantly impact the whole field of drug development. The results presented here can significantly accelerate the translation into the clinics of known compounds for novel therapeutic uses.

A review of connectivity map and computational approaches in pharmacogenomics

“…There are also many other functional phenotype-based approaches that use the CMap resource to understand MoA [7,[78][79][80]. It is widely known that many drugs with therapeutic targets in cancer prognosis and diagnosis have been identified using CMap.…”

“…the Connectivity Map (CMap) approach [3][4][5]. However, for either type of investigations, machine learning [6] and biomedical text mining [7] approaches have been vital to uncover hidden relationships between drugs and potential new indications. Overall, applying these methods on drug perturbation data sets has proven to be beneficial in enhancing the understanding of the connection between genes, drugs and diseases [8][9][10] because such methodologies can lead to generation of novel hypotheses beyond classical pharmacology by translating new knowledge from genomic in vitro screens and cell-based assays to the patients.…”

Integrative methods for analyzing big data in precision medicine

“…After combining these kernel matrices into a single kernel matrix, the authors applied a Support Vector Machine (SVM), a supervised machine learning method for classification. They trained the SVM on the existing drug classification achieving 78% of classification accuracy and they used the top scoring misclassified drugs as new candidates for repurposing [138]. A similar approach was used by Wang et al [139], who developed a PreDR (Predict Drug Repurposing) method where drug-centered kernel matrices represent: 1) drug chemical similarities obtained from PubChem database; 2) target (protein) sequence similarities retrieved from KEGG BRITE and DrugBank; and 3) drug side-effect similarities for SIDER database.…”

“…Joint kernel matrices [138] Drug repurposing by integrating of drug chemical structures, PPI network and drug induced gene expression data.…”

Drug Repurposing by Simulating Flow Through Protein–Protein Interaction Networks

“…The use of marketed drugs for new indications, referred to as drug repurposing, can significantly decrease expenses 1. Algorithms used for drug repurposing analyze several kinds of information either individually or in combination, including gene expression data, chemical properties of substances, drug target data, and static properties of intracellular networks 2, 3. Networks are reliable frameworks for the simulation of biological processes.…”