Exploring structural requirements of HDAC10 inhibitors through comparative machine learning approaches

Area of Science:

• Pharmacology
• Medicinal Chemistry
• Computational Biology

Background:

• Histone deacetylase (HDAC) inhibitors are key in anticancer drug development.
• HDAC10, a class-IIb HDAC, plays a role in cancer progression.
• The lack of human HDAC10 structural data hinders structure-based drug design.

Purpose of the Study:

• To develop ligand-based modeling techniques for identifying selective HDAC10 inhibitors.
• To accelerate the design of novel HDAC10-targeted anticancer agents.
• To provide insights for medicinal chemists in developing efficient HDAC10 inhibitors.

Main Methods:

• Applied diverse ligand-based modeling techniques on 484 HDAC10 inhibitors.
• Developed machine learning (ML) models for compound screening.
• Utilized Bayesian classification and Recursive partitioning for structural analysis.
• Performed molecular docking to understand binding patterns.

Main Results:

• Successfully developed ML models capable of screening potential HDAC10 inhibitors.
• Identified key structural fingerprints associated with HDAC10 inhibitory activity.
• Gained insights into the binding interactions within the HDAC10 active site.

Conclusions:

• Ligand-based modeling offers a viable approach for HDAC10 inhibitor design in the absence of structural information.
• The developed models can facilitate the screening of large chemical databases for novel HDAC10 inhibitors.
• This study provides valuable information for the rational design of targeted anticancer therapies.