One-dimensional molecular representations and similarity calculations: methodology and validation

Area of Science:

• Computational Chemistry
• Medicinal Chemistry
• Drug Discovery

Background:

• Massive biological screening dominates drug discovery, often overlooking computational methods.
• Existing 2D and 3D structure representations have limitations in high-throughput screening.
• Simpler methods are needed to enhance the rationality of early drug discovery stages.

Purpose of the Study:

• To introduce and validate a novel 1D representation for chemical structures.
• To assess the performance of 1D representations against established 2D and 3D methods.
• To explore a simpler approach for structure comparison in drug discovery.

Main Methods:

• Generated 1D representations by collapsing 3D molecular models or 2D graphs onto a single axis.
• Differentiated atoms based on element type, hybridization, and connectivity.
• Aligned 1D representations to calculate structural similarity and validated against structure-activity data.

Main Results:

• 1D similarity measures consistently outperformed Daylight 2D fingerprints.
• 1D similarity measures demonstrated superior performance compared to Cerius(2) pharmacophore fingerprints.
• The 1D method proved effective in structure-activity relationship validation tests.

Conclusions:

• The novel 1D chemical structure representation offers a significant advantage over existing methods.
• This simplified approach enhances the ability to represent and compare chemical structures.
• 1D representations present a promising, computationally efficient tool for drug discovery.