SpaceHASTEN: A Structure-Based Virtual Screening Tool for Nonenumerated Virtual Chemical Libraries.

Area of Science:

• Computational chemistry
• Drug discovery
• cheminformatics

Background:

• Fully enumerating chemical libraries for drug discovery is computationally infeasible due to vast chemical space.
• Non-enumerated virtual chemical spaces offer a practical alternative, representing compounds as building blocks connected by rules.
• Existing tools for searching these spaces primarily use ligand-based methods, lacking protein structure-based screening capabilities.

Purpose of the Study:

• To develop a novel hybrid ligand/structure-based virtual screening tool, SpaceHASTEN.
• To enable efficient structure-based virtual screening of non-enumerated chemical spaces using protein structure as input.
• To address the gap in software for screening non-enumerated chemical spaces with protein structure queries.

Main Methods:

• Development of SpaceHASTEN, integrating SpaceLight, FTrees, LigPrep, and Glide.
• Utilizing a hybrid approach combining ligand-based and structure-based virtual screening methods.
• Validation using three public targets from the DUD-E dataset.

Main Results:

• SpaceHASTEN successfully performed structure-based virtual screening on non-enumerated chemical spaces.
• The tool identified a substantial number of diverse and novel high-scoring compounds (virtual hits).
• Screening millions of compounds allowed retrieval of hits from billions of molecules in the chemical space.

Conclusions:

• SpaceHASTEN provides an efficient solution for structure-based virtual screening of large, non-enumerated chemical spaces.
• The developed tool facilitates the discovery of novel drug candidates by leveraging protein structure information.
• The freely available software offers a valuable resource for computational drug discovery efforts.