Rchemcpp

Rchemcpp identifies structurally similar compounds (structural analogs) within large molecular databases to support compound prioritization and toxicology analyses.

Key Features:

• Molecule Kernels: Employs state-of-the-art molecule kernels for structural similarity assessment, providing superior performance compared with traditional similarity measures and strong utility in machine learning applications.
• Efficient Prefiltering Strategy: Implements an efficient prefiltering strategy that reduces computational cost while preserving sensitivity of similarity assessments.
• Database Integration: Supports querying of large molecular repositories including ChEMBL, DrugBank, and the Connectivity Map for analog retrieval.

Scientific Applications:

• Compound Prioritization: Enables prioritization of compounds after high-throughput screening based on structural similarity to known actives.
• Toxicology and Safety Assessment: Assists in identifying analogs to anticipate and minimize adverse side effects during later stages of drug development.
• Machine Learning Pipelines: Has been integrated into the DeepTox pipeline and contributed to success in the Tox21 Data Challenge.

Methodology:

Computational methods explicitly include molecule kernels for similarity scoring, an efficient prefiltering strategy to reduce computations, and querying of ChEMBL, DrugBank, and the Connectivity Map.