Drug-Target Interaction

Overview

The drug-target-interaction dataset is a combinatorial complex representing drug-target interactions and similarity relationships. The dataset is based on the work by Perlman et al., which combines multiple drug and gene similarity measures to predict drug-target interactions.

Structure

The complex has three cell levels:

• 0-cells: drugs, identified by DrugBank IDs, and protein targets, identified by Entrez Gene IDs.
• 1-cells: drug-target interaction pairs, indicating potential or known interactions between drugs and protein targets.
• 2-cells: similarity groups based on drug similarity measures (ATC hierarchy, chemical structure, ligand Jaccard, CMap Jaccard, and side effects) and target similarity measures (distance, Gene Ontology, and sequence similarity).

This combines direct drug-target interactions with higher-order similarity relationships.

Statistics

0-cells (Nodes)

565

Node Types

Drugs and Protein Targets

Interactions

79,627

Interaction Types

1-cells and 2-cells

2-cells (Faces)

877

Face Type

Similarity Groups

Applications

This structure is useful for:

• Drug repositioning based on similarity patterns.
• Protein target prediction for drugs.
• Polypharmacology and network pharmacology studies.
• Machine learning models that use topological features for drug-target interaction prediction.

Data Source

The original data comes from Perlman et al.’s study on combining drug and gene similarity measures. The dataset includes interaction data from DrugBank and protein target information from Entrez Gene, integrated with multiple similarity measures to create a comprehensive representation of the drug-target interaction landscape.