Computational Identification of Chemical Compounds with Potential Activity against Leishmania amazonensis using...

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Machine learning in drug discovery

Background:

• Leishmaniasis is a neglected tropical disease affecting 98 countries, with increasing morbidity and mortality.
• Current treatments for leishmaniasis have significant drawbacks, including toxicity, high cost, and inconvenient administration routes.
• There is a critical need for novel, effective, and accessible treatments for leishmaniasis.

Purpose of the Study:

• To develop computational models for identifying novel chemical compounds with anti-leishmanial activity.
• To leverage machine learning techniques for predicting the efficacy of compounds against Leishmania amazonensis.
• To facilitate the discovery of new drug candidates for leishmaniasis treatment.

Main Methods:

• Utilized a dataset of 116 organic chemicals with known activity against Leishmania amazonensis promastigotes (IC50 ≤ 1.5μM).
• Calculated molecular descriptors using Dragon software.
• Developed and validated machine learning models (k-nearest neighbors, classification trees, artificial neural networks, support vector machine) using WEKA.
• Performed virtual screening of chemical libraries to identify potential anti-leishmanial agents.

Main Results:

• Machine learning models achieved high accuracy (82-91%) on the training set.
• Models demonstrated excellent sensitivity (97-100%) and specificity (92-94%).
• Virtual screening identified 156 compounds as potential anti-leishmanial agents.

Conclusions:

• Machine learning-based techniques are effective for discovering new anti-leishmanial compounds.
• The developed models provide a valuable tool for accelerating drug discovery for leishmaniasis.
• This approach offers a promising alternative to traditional drug discovery methods for neglected tropical diseases.