TropMol-Caipora: A Cloud-Based Web Tool to Predict Cruzain Inhibitors by Machine Learning

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and development
• Parasitology and infectious diseases

Background:

• Chagas disease (CD) is a Neglected Tropical Disease affecting millions, with limited treatment options.
• Cruzain, a cysteine protease from *T. cruzi*, is a validated therapeutic target for CD.
• Development of novel cruzain inhibitors is crucial for effective CD treatment.

Purpose of the Study:

• To develop a publicly accessible, free online molecular screening model for predicting cruzain inhibitors.
• To identify key molecular descriptors associated with cruzain inhibitory activity.
• To facilitate rapid screening of potential drug candidates for Chagas disease.

Main Methods:

• Utilized a Random Forest model for quantitative structure-activity relationship (QSAR) analysis.
• Trained the model on a dataset of approximately 8,000 compounds with over a million calculated descriptors.
• Validated model performance using R-squared and Root Mean Squared Error (RMSE) metrics on training and test sets, and through 5-fold cross-validation.

Main Results:

• The Random Forest model demonstrated high predictive accuracy with R-squared values of 0.91 (training) and 0.72 (test).
• Key molecular features influencing cruzain inhibition were identified: aromaticity (especially nitrogenous rings), halogenation, molecular accessibility, and charge.
• Rigid or bicyclic structures were found to potentially decrease inhibitory efficiency.

Conclusions:

• The developed screening model provides a valuable, free resource for identifying potential cruzain inhibitors.
• Aromaticity and halogenation are significant positive contributors to cruzain inhibitory activity.
• Understanding structure-activity relationships guides the design of more effective Chagas disease therapeutics.