Graph-based similarity concepts in virtual screening.

Area of Science:

• Medicinal Chemistry
• Computational Drug Design

Background:

• Drug design relies on identifying novel compounds with desired properties.
• Quantifying molecular similarity is essential for predicting compound efficacy but remains challenging.
• Numerous computational approaches exist for assessing molecular similarity.

Purpose of the Study:

• To review general aspects and recent developments in molecular similarity criteria for drug design.
• To discuss various computational methods for encoding molecules for similarity comparisons.
• To highlight advanced approaches considering chemical modifications in existing drugs.

Main Methods:

• Review of existing literature on molecular similarity metrics.
• Discussion of computational techniques including 3D structure, fingerprints, graph-based, and alignment-based methods.
• Analysis of approaches that consider chemical modifications across different therapeutic categories.

Main Results:

• Similarity quantification is a key challenge in drug discovery.
• Various computational methods enable molecule comparison for virtual screening.
• Advanced techniques incorporate chemical modifications found in marketed drugs.

Conclusions:

• Effective molecular similarity assessment is vital for efficient drug design.
• A range of computational tools, from fingerprints to 3D structures, aid in this assessment.
• Future directions may involve deeper analysis of chemical modifications for improved compound identification.