GIFT

GIFT infers chemogenomic features from drug–target interactions using a global optimization framework to reveal substructure–domain relationships for fragment-based drug discovery and drug repositioning.

Key Features:

• Implementation (C++): GIFT is provided as a C++ package for computational inference and analysis of chemogenomic features.
• Global Optimization Approach: Employs a global optimization strategy to assess substructure–domain interactions simultaneously, mitigating the risk of high false-positive estimations.
• Combinatorial Analysis: Incorporates combinations of chemical substructures to evaluate their joint contribution to drug–target interactions.
• Predictive Performance: Accurately predicted 81% of known drug–domain interactions in a set of 53 examples and, among the top 100 predicted combined substructure–domain interactions, identified 18 with corresponding structures in the Protein Data Bank, 15 of which were experimentally validated.
• Drug Repositioning Findings: Predicted anticancer activities for tazarotene, adapalene, acitretin, and raloxifene.

Scientific Applications:

• Chemogenomic feature inference: Infers associations between synthetic drug substructures and protein domains.
• Fragment-based drug discovery: Supports analysis of fragment–domain interactions relevant to fragment-based design.
• Drug repositioning and repurposing: Prioritizes existing drugs for potential new therapeutic indications based on chemogenomic profiles.
• Structural validation mapping: Enables mapping predicted interactions to Protein Data Bank structures for experimental validation.

Methodology:

Implemented in C++, GIFT uses a global optimization strategy and combinatorial analysis of chemical substructures to infer substructure–domain interactions and was evaluated against known drug–domain interactions and Protein Data Bank structural mappings.