RISC: Rapid Inverted-Index Based Search of Chemical Fingerprints.

Area of Science:

• Chemoinformatics
• Computational Chemistry
• Drug Discovery

Background:

• Searching large molecular databases is crucial for drug discovery.
• High-dimensional and sparse chemical fingerprints (e.g., unfolded fingerprints) are commonly used.
• Existing search methods struggle with the scale and sparsity of these fingerprints.

Purpose of the Study:

• To develop a novel, efficient searching algorithm for massive molecular repositories.
• To address the challenges posed by high-dimensional and sparse chemical fingerprints.
• To significantly accelerate searching efficiency in chemoinformatics.

Main Methods:

• Proposed a new algorithm named RISC (Rapid Information Search and Caching).
• RISC leverages the sparsity of high-dimensional fingerprints for effective pruning.
• Algorithm tested on both binary and non-binary chemical fingerprints.

Main Results:

• RISC demonstrates significant speed-up in searching efficiency compared to state-of-the-art techniques.
• Experiments show consistent performance gains on Range Queries and Top-k Queries.
• Effectiveness proven for fingerprints with dimensions of 2048 and above, common in unfolded fingerprints.

Conclusions:

• RISC offers a robust and efficient solution for searching large molecular datasets.
• The algorithm's ability to exploit sparsity dramatically improves search performance.
• This advancement has substantial implications for accelerating chemoinformatics and drug discovery pipelines.