Use of reduced graphs to encode bioisosterism for similarity-based virtual screening

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Similarity-based virtual screening is crucial for identifying potential drug candidates.
• Bioisosteric fragment equivalences represent structural modifications that maintain biological activity.
• Encoding these equivalences can enhance virtual screening but requires careful implementation.

Purpose of the Study:

• To investigate the inclusion of explicit bioisosteric fragment equivalence information in similarity-based virtual screening.
• To assess the impact of this information on identifying novel chemical scaffolds and potential drug leads.

Main Methods:

• Utilized reduced graphs to encode known bioisosteric equivalences from the BIOSTER database.
• Performed scaffold-hopping experiments using the WOMBAT database to evaluate the effectiveness of the encoding method.

Main Results:

• Incorporating bioisosteric information successfully identified similarities between reference and active structures with different yet equivalent fragments.
• However, these equivalences also led to increased similarity with inactive compounds, potentially overwhelming true positives.

Conclusions:

• Explicitly encoding bioisosteric fragment equivalences can aid in discovering novel chemical entities in virtual screening.
• The challenge lies in mitigating the confounding effect of bioisosteric similarities with inactive compounds for reliable virtual screening.