Computational identification of chemical compounds with potential anti-Chagas activity using a classification tree

Area of Science:

• Computational chemistry
• Drug discovery
• Parasitic diseases

Background:

• Chagas disease is a significant public health concern in Latin America, affecting millions.
• Current treatments for Chagas disease are limited and often unsatisfactory.
• There is a critical need for novel therapeutic agents to combat this neglected tropical disease.

Purpose of the Study:

• To develop and validate a computational model for identifying novel compounds with potential anti-chagasic activity.
• To leverage a large dataset of chemical compounds for drug discovery.
• To demonstrate the efficacy of a rational, computer-based approach in accelerating anti-chagasic drug development.

Main Methods:

• Utilized a dataset of 584 compounds from the Drugs for Neglected Diseases initiative.
• Employed Dragon software for molecular descriptor calculation.
• Developed a classification tree model using WEKA software for compound screening.
• Validated the model using 10-fold cross-validation and an independent test set.
• Performed simulated ligand-based virtual screening for promising anti-chagasic agents.

Main Results:

• The computational model achieved high accuracy (over 93.4% on the training set, over 90.5% and 92.2% on validation and test sets).
• Sensitivity and specificity values consistently remained around 90%.
• False alarm rates were below 10.5% across all validation sets.
• Virtual screening results showed good agreement with existing experimental data for known anti-chagasic compounds.

Conclusions:

• The developed computational model is a highly accurate and reliable tool for identifying potential anti-chagasic compounds.
• This rational, computer-based drug discovery method can significantly reduce costs and expedite the development of new treatments for Chagas disease.
• The approach offers a promising strategy for addressing neglected tropical diseases like Chagas disease.