Molecular similarity: a key technique in molecular informatics

Area of Science:

• Computational Chemistry
• Cheminformatics
• Drug Discovery

Background:

• Molecular Informatics leverages concepts like similarity to identify relationships between molecules.
• Molecular similarity is extensively used in drug discovery for lead discovery and compound optimization.
• Pharmaceutical companies widely employ similarity methods for virtual screening and predicting ADME/Tox and physicochemical properties.

Purpose of the Study:

• To discuss molecular structure representation (descriptors) and comparison methods.
• To explore the concept of molecular similarity, its definitions, uses, and evolution.
• To evaluate and challenge accepted views on molecular similarity, especially regarding biological receptor interactions.

Main Methods:

• Discussion of molecular descriptors for structure representation.
• Analysis of methods for comparing molecular structures.
• Evaluation of the relationship between molecular comparisons and measured properties.

Main Results:

• Molecular similarity has practical importance and widespread applications in the pharmaceutical industry.
• The concept of molecular similarity contains implicit and explicit assumptions.
• The context of molecular similarity is critical in defining and limiting its use.

Conclusions:

• Molecular similarity's utility is context-dependent, influencing predictions of molecular binding and efficacy.
• Key issues include solvation effects, binding site heterogeneity, and the choice of similarity measures.
• A critical evaluation of molecular similarity is necessary for its effective application in drug discovery.