DrugCombDB: a comprehensive database of drug combinations toward network medicine and combination therapy

Abstract

Drug combinations have demonstrated high efficacy and low adverse side effects compared to single drug administrations in cancer therapies, and thus draw intensive attentions from researchers and pharmaceutical enterprises. Thanks to the fast development of high-throughput screening (HTS) methods, the amount of available drug combination datasets has tremendously increased. However, existing drug combination databases are lack of indications of the drug combinations and quantitative dose-responses. Therefore, there is an urgent need for a comprehensive database that is crucial to both experimental and computational screening of drug combinations. In this paper, we present DrugCombDB, a comprehensive database dedicated to integrating drug combinations from various data sources. Concretely, the data sources include 1) high-throughput screening assays of drug combinations, 2) external databases, and 3) manual curations from PubMed literature. In total, DrugCombDB includes 1,127,969 experimental data points with quantitative dose response and concentrations of drug combinations covering 561 unique drugs and 104 human cancer cell lines, and 1,875 FDA approved or literature-supported drug combinations. In particular, we adopted the zero interaction potency (ZIP) model [2] to compute the scores determining the synergy or antagonism of two drugs. To facilitate the downstream usage of our data resource, we prepared multiple datasets that are ready for building prediction models of classification and regression analysis. A website with user-friendly data visualization is provided to help users access the wealth of data. Users can input a drug of interest to retrieve associated drug combinations, together with the supporting evidence sources and drug targets. Our database is available at http://drugcombdb.denglab.org/.