Design, synthesis and experimental validation of novel potential chemopreventive agents using random forest and...

Area of Science:

• Computational chemistry
• Drug discovery
• Cancer chemoprevention

Background:

• Limited knowledge exists on organic compounds for cancer prevention.
• Need for novel chemopreventive agents with minimal toxicity.

Purpose of the Study:

• To develop predictive computational models for identifying chemopreventive agents.
• To virtually screen natural products for cancer chemoprevention potential.

Main Methods:

• Curated a database of over 400 compounds with known chemoprevention activities.
• Developed and validated random forest and support vector machine models.
• Virtually screened a library of ~23,000 natural products and derivatives.

Main Results:

• Identified 148 novel potential chemopreventive compounds via consensus prediction.
• Synthesized and experimentally validated 18 compounds.
• Experimental results confirmed model predictions, demonstrating utility.

Conclusions:

• Developed predictive models are effective for identifying novel chemopreventive lead compounds.
• Computational screening accelerates the discovery of low-toxicity cancer prevention agents.
• Models can be applied to screen additional chemical libraries for chemoprevention discovery.