Sachem

Sachem implements scalable substructure search and fingerprint-based indexing within PostgreSQL to enable efficient querying of small-molecule databases for chemical informatics applications.

Key Features:

• Substructure Search Methods: Provides two methods: a performance-oriented reimplementation of substructure indexing based on the OrChem fingerprint, and a novel approach using newly designed fingerprints stored in inverted indices.
• Fingerprint-based Screening: Uses fingerprint-based screening to reduce the number of verification calls during substructure searches.
• Inverted Index Storage: Stores newly designed fingerprints in inverted indices to enhance retrieval efficiency and scalability.
• Performance and Scaling: Demonstrates sublinear performance scaling with dataset size and superior scaling compared to existing freely available cartridges, validated on datasets from 1 million to 94 million compounds.
• Screen-out Efficiency: Optimizes screen-out efficiency by querying and processing large amounts of pre-extracted fingerprint information to limit expensive verification procedures.

Scientific Applications:

• Drug Discovery: Facilitates rapid identification of compounds containing desired substructures to support lead identification and optimization.
• Chemical Informatics: Enables scalable indexing and querying of chemical databases for medicinal chemistry and cheminformatics analyses.
• Toxicological Studies: Assists screening of large chemical libraries for structural motifs associated with toxicity.

Methodology:

Implements a reimplementation of OrChem fingerprint-based substructure indexing and a novel inverted-index method using newly designed fingerprints; employs fingerprint-based screening to reduce verification calls and uses inverted indices to optimize retrieval.