Average Information Content Maximization--A New Approach for Fingerprint Hybridization and Reduction

Area of Science:

• Cheminformatics and Computational Chemistry
• Drug Discovery and Development

Background:

• Chemical fingerprints are crucial for representing molecular structures in cheminformatics.
• High-resolution fingerprints, while effective, contain noise from irrelevant bits, increasing computational time.
• Efficient representation is key for large-scale virtual screening and drug design.

Purpose of the Study:

• To develop a novel method for constructing hybrid reduced chemical fingerprints.
• To identify and select the most informative bits from existing fingerprint representations.
• To improve the efficiency and performance of virtual screening campaigns.

Main Methods:

• Introduction of the Average Information Content Maximization (AIC-Max) algorithm.
• Application of AIC-Max to construct reduced fingerprints from ligands of five serotonin receptors (5-HT2A, 5-HT2B, 5-HT2C, 5-HT5A, 5-HT6).
• Utilizing 100 selected bits from four non-hashed fingerprints for in silico discrimination.

Main Results:

• The AIC-Max algorithm successfully identified a concise set of 100 informative bits.
• These reduced fingerprints captured essential structural information for discriminating receptor ligands.
• Classification experiments showed the reduced representation achieved performance comparable to, or slightly better than, longer fingerprints.

Conclusions:

• Hybrid reduced fingerprints constructed using AIC-Max offer a highly efficient representation.
• This method significantly reduces computational time in virtual screening without compromising accuracy.
• The AIC-Max algorithm presents a valuable tool for optimizing cheminformatics workflows in drug discovery.