Epigenetic Target Fishing with Accurate Machine Learning Models.

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Epigenetic drugs are crucial for cancer treatment, with eight approved therapies.
• Vast chemogenomic data exists for epigenetic targets but remains underexploited for predictive modeling.
• Structure-activity relationships in epigenetic data can inform medicinal chemistry.

Purpose of the Study:

• To develop accurate predictive models for small molecule epigenetic target profiling.
• To leverage large-scale chemogenomic data for drug discovery.
• To create a tool supporting medicinal chemistry efforts in identifying epigenetic modulators.

Main Methods:

• Analyzed 26,318 compounds with quantitative biological activity data for 55 epigenetic targets.
• Systematically compared machine learning models trained on various molecular fingerprints.
• Developed and validated predictive models for epigenetic target identification.

Main Results:

• Achieved high accuracy in predicting small molecules' epigenetic target profiles.
• Validated models demonstrated mean precisions up to 0.952 for epigenetic target prediction.
• Identified significant potential for the models to discover novel epigenetic modulators.

Conclusions:

• The developed predictive models show considerable potential for identifying small molecules with epigenetic activity.
• The findings support the advancement of drug discovery for epigenetic therapies.
• A freely accessible web application was created to implement these results.