Similarity searching in databases of flexible 3D structures using smoothed bounded distance matrices

Area of Science:

• Computational chemistry
• Cheminformatics
• Molecular modeling

Background:

• Chemical structure similarity searching is crucial for drug discovery.
• Existing 2D methods often fail to capture 3D conformational nuances.
• Representing molecules as 3D graphs offers a richer structural description.

Purpose of the Study:

• To develop and evaluate a novel 3D molecular graph-based similarity method.
• To assess the method's efficacy in retrieving compounds with similar biological activity.
• To compare the performance against established 2D similarity searching techniques.

Main Methods:

• Utilizing a graph matching procedure for comparing 3D molecular graphs.
• Incorporating conformational flexibility using distance ranges between atom pairs.
• Employing triangle and tetrangle bound smoothing from distance geometry to generate distance ranges.

Main Results:

• The proposed 3D method demonstrates effectiveness in retrieving biologically similar compounds.
• Performance evaluation shows superiority over traditional 2D similarity searching approaches.
• The method successfully accounts for conformational variations in chemical structures.

Conclusions:

• The 3D molecular graph similarity method offers an improved approach for virtual screening.
• Accounting for conformational flexibility is key to accurate biological activity prediction.
• This method enhances the discovery of novel drug candidates with desired properties.