1 - 44 of 44 results


Extracts molecular fragments, classified as bricks and linkers, from small molecule datasets. eMolFrag uses the fragments in order to construct targeted libraries for virtual screening. It stores the connectivity information for the extracted building blocks to help generate new series of chemically feasible compounds. It is optimized to work with eSynth, a recently developed molecular synthesis algorithm. It can also be integrated into other cheminformatics toolkits utilizing chemical fragments.


A dual-regularized one class collaborative filtering algorithm for biological relation prediction. REMAP explores continuous chemical space, protein space, and their interactome on a large scale. When tested in a reliable, extensive, and cross-gene family benchmark, REMAP outperforms the state-of-the-art methods. It can screen a dataset of 200 thousand chemicals against 20 thousand proteins within 2 hours. Using the reconstructed genome-wide target profile as the fingerprint of a chemical compound, we predicted that seven FDA-approved drugs can be repurposed as novel anti-cancer therapies. The anti-cancer activity of six of them is supported by experimental evidences. Thus, REMAP is a valuable addition to the existing in silico toolbox for drug target identification, drug repurposing, phenotypic screening, and side effect prediction.

LRSSL / Laplacian regularized sparse subspace learning

Predicts and interprets drug–disease associations of new drugs and approved drugs. LRSSL integrates drug chemical information, drug target domain information and target annotation information. It outperforms several recent approaches for predicting drug–disease associations. The tool could be easily extended to include more feature profiles and similarity metrics. It can extract drug features from different types of feature profiles.

MANTRA / Mode of Action by NeTwoRk Analysis

Allows to analyze the Mode of Action (MoA) of novel drugs and to identify known and approved candidates for “drug repositioning”. MANTRA is a computational tool based on network theory and non-parametric statistics on gene expression data, which offers three “workspaces”: Analysis, Network and Search. Users can visually explore the Drug Network that provides, for each of the drugs, information about biochemical interactions, therapeutic indications, known MoA, pharmacology and targeted proteins.


An integrated data driven approach to drug repositioning. GeneDiseaseRepositioning is an exhaustive approach for identifying new uses for existing drugs, with a focus on gene-disease (G-D) associations. A Bayesian statistics approach was applied, as means of integrating and ranking G-D associations captured in 10 primary data sources. Scored G-D associations are then integrated with other biological entities to produce a semantic network for target-driven drug repositioning. A method for the automated detection of therapeutic areas of interest is also introduced. Finally, a four node semantic subgraph was introduced, and mine the integrated network for instances of this subgraph, using an algorithm. Novel drug-disease interactions inferred from the network are then ranked. It is expected that this approach will facilitate further research on drug repositioning.

DPDR-CPI / Drug Candidate Positioning and Drug Repositioning via Chemical-Protein Interactome

A server which can make real-time predictions based only on the structure of the small molecule. DPDR-CPI is able to produce indication predictions for a user molecule towards ~1,000 human diseases, providing suggestions for drug candidate positioning and drug repositioning. It has the potential to improve the drug development pipeline in terms of indication prioritization. It achieves a reasonably good overall performance and can be utilized for drug candidate positioning and repositioning purposes.


A novel computational method to identify potential novel indications for a given drug. MBiRW utilizes some comprehensive similarity measures and Bi-Random walk algorithm. Drug similarity network and disease similarity network are firstly constructed, and they are incorporated into a heterogeneous network with known drug-disease interactions. Based on the drug-disease heterogeneous network, Bi-Random walk algorithm predicts novel potential drug-disease associations. Computational experiment results from various datasets demonstrate that the proposed approach has reliable prediction performance and outperforms several recent computational drug repositioning approaches.

Drug voyager

Proposes a platform to construct a drug-specific pathway in which a molecular-level mechanism of action is formulated based on pharmacologic, pharmacogenomic, transcriptomic, and phenotypic data related to drug response. With the Drug voyager platform, the molecular-level action of a drug is represented by connecting the three conceptual levels of “initiation,” “perturbation,” and “destination.” Each level includes a combination of the five types of seed genes related to drug responses and phenotypes: drug target genes (TG), pharmacogenomic variant genes (VG), differentially expressed genes (DEG), disease genes (DisG), and side-effect genes (SEG). As a consequence of construction of level-to-level pathways, 82 drug-signaling pathways were generated in total for 82 drugs. In the validation step, these pathways were significantly enriched in known drug pathway databases and show higher significance levels compared to when other models are used.

Cogena / Co-expressed gene-set enrichment analysis

A fully configurable framework for co-expressed gene set enrichment analysis. By combining pathway analysis and drug repositioning analysis, Cogena provides a unique approach to imply the drug mode of action in a disease context, which is important to the translational development of computationally repositioned drugs. Cogena is a powerful tool for co-expressed gene set enrichment analysis, including pathway analysis and drug repositioning.

miRDDCR / miRNA-based method to extensively predict Drug-Disease Causal Relationships

Predicts drug-disease causal relationships. miRDDCR is a miRNA-based method that reveals the causal relationships between drugs and diseases under their molecular basis miRNAs. The method relies on the hypothesis that similar small molecules tend to target similar miRNAs, and finally treats similar diseases. It allows prediction of drug-disease relationships in a large scale by combining similarity measurements, existing drug-miRNA associations and miRNA-disease associations.

Drug Repurposing Hub

Aims to discover new indications for existing drugs with known safety profiles. Drug Repurposing Hub is an interactive website designed to rapidly identify drugs for evaluation in disease models. The Hub consists of (i) a physical drug screening library available in multiple plate formats at the Broad Compound Management facility, (ii) manually curated annotations as part of a comprehensive publicly-accessible information resource with data API, and (iii) experimental results with liquid chromatography-mass spectrometry (LC-MS) tracings and future cellular assays.


Enables the discovery of putative leads for drug repositioning that can be applied to any microbial proteome, as long as the interactome of interest is at least partially known. DTIPRED can be used in the identification of new leads for drug repositioning. It predicts the probability of each DTI pair to interact the probability of a given DTI pair to be classified as a positive interaction. The tool uses a classification model based on random frests of decision trees which can solve complex classification problems in large data sets with a significant number of features.

PDOD / Prediction of Drugs having Opposite effects on Disease genes

Identifies drugs having opposite effects on altered states of disease genes. PDOD proposes a scoring function to discover drugs likely to restore altered states of disease genes using the path from a drug to a disease through the drug-drug target interactions, shortest paths from drug targets to disease genes in molecular pathways, and disease gene-disease associations. The method provides a simple web service that researchers can submit genes of interest with their altered states and will obtain drugs seeming to have opposite effects on altered states of input genes.


An algorithm for the large-scale prediction of drug indications, that can handle both approved drugs and novel molecules. PREDICT is based on the observation that similar drugs are indicated for similar diseases, and utilizes multiple drug–drug and disease–disease similarity measures for the prediction task. Our method attained high rates of specificity and sensitivity in cross-validation (AUC=0.9), surpassing existing methods. Furthermore, our predictions attained significant coverage of drug–disease associations tested in clinical trials and are in good agreement with tissue-specific expression information on the targets of the corresponding drugs, suggesting that they can be regarded as valuable leads for further research.


Hosts a knowledgebase designed for drug discovery. The Integrity database contains a large collection of drugs which are annotated with information on their respective drug targets, the diseases they are associated with, and the clinical phases of the drugs. Drug targets are assigned a status in Integrity, which can be ‘Validated’, ‘Candidate’, ‘Exploratory’, or none. Validated drug targets are associated with drugs under active development in clinical phases or with launched drugs for the disease of interest. Candidate drug targets are associated with drugs that are no longer under active development for the respective disease. Exploratory drug targets are associated with drugs that are currently under biological investigation for the disease. In Integrity, drugs are not directly linked to genes. Instead, drugs are linked to internal target IDs and these targets are then linked to Entrez Gene identifiers.

MeSHDD / MeSH-based Drug-Drug Similarity and Repositioning

A framework for computational drug repositioning using literature-derived drug-drug similarity. MeSHDD provides an alternate way of searching the biomedical corpus for novel (and existing) uses of approved drugs. It expanded previous methods using curated MeSH terms from MEDLINE to find drug-MeSH term pairs that were enriched for co-occurrence in the medical literature and developed a method for calculating pair-wise similarities between drugs. MeSHDD is provided as open-source code and deployed as a free-to-use, interactive application to explore the database of similarity-based drug clusters.


An integrative network encoding knowledge from millions of biomedical studies. Hetionet v1.0 consists of 47031 nodes of 11 types and 2250197 relationships of 24 types. Data was integrated from 29 public resources to connect compounds, diseases, genes, anatomies, pathways, biological processes, molecular functions, cellular components, pharmacologic classes, side effects, and symptoms. Hetionet was created for Project Rephetio, which aims to systematically identify why drugs work and predict new therapies for drugs. Hetionet network is accessible via a Neo4j Browser and is available for download in three formats: JSON, Neo4j and TSV.

Project Rephetio Browser

Predicts new uses for existing compounds. Project Rephetio Browser uses machine learning to systematically learn network patterns of drug efficacy. This method translates the network paths between a compound and disease into a predicted probability of treatment. It makes predictions for 1538 approved small molecule compounds and 136 complex diseases, resulting in a total of 209168 compound-disease pairs. Predictions are created from Hetionet v1.0, an integrative network of biomedicine that contains 2250197 relationships of 24 types. Data about compounds (identifiers, names, and desciptions) are from DrugBank, while diseases are from the Disease Ontology.

RE:fine drugs

An interactive user interface to integrate GWAS and PheWAS reposition datasets using Drug–Gene–Disease triads along with advanced search and export capabilities. ‘RE:fine drugs’ enables researchers to explore dataset of drug repurposing pairs to discover novel opportunities for possible treatment of new indications. The proposed usage of ‘RE:fine Drugs’ is under a hypothesis generation framework rather than a statistical testing framework: a positive identification from this database only suggests a possibility of drug reposition, rather than any statistical validation.


A package developed for the large-scale analysis of gene expression signatures. GeneExpressionSignature implements two rank-merging algorithms and two similarity-scoring algorithms. It provides a flexible solution for gene expression signature-based studies and holds great potential in biomedical research applications, such as drug repurposing. All of the functions in the GeneExpressionSignature package, except getRLs, support ratio, log-ratio, and rank data stored as assay data in the ‘‘ExpressionSet’’ object of the Biobase package as input data. The label of each column, as well as phenotypic data in the ‘‘ExpressionSet’’ object, is the biological state descriptions of the gene expression profiles.

DR. PRODIS / DRugome PROteome and DISeasome

A comprehensive proteome scale approach that predicts human protein targets and side effects of drugs. We develop a comprehensive proteome scale approach that predicts human protein targets and side effects of drugs. For drug-protein interaction prediction, FINDSITEcomb, whose average precision is ~30% and recall ~27%, is employed. For side effect prediction, a new method is developed with a precision of ~57% and a recall of ~24%. DR. PRODIS provides protein targets of drugs, drugs for a given protein target, associated diseases and side effects of drugs, as well as an interface for the virtual target screening of new compounds.

DMAP / Drug directionality Map

An in silico drug-protein connectivity map, which contains directed drug-to-protein effects and effect scores. DMAP that can help drug development researchers evaluate what effects a drug may have on disease-relevant genes or proteins. DMAP compiles each drug's stimulatory or inhibitory effects on genes or their protein products, based on the computational integration of such data from different databases. It covers 438,004 chemical-to-protein effect relationships between 24,121 PubChem compounds that cover 289,571 chemical entities with a synonymous name, and 5,196 distinct UniProt proteins. DMAP may be used wherever CMAP data coverage is poor for drug repositioning applications. We demonstrate that DMAP can successfully recall known drugs for examined disease indications.