Abstract:
Leishmania, categorized as a neglected tropical ailment, is instigated by a protozoan belonging to the leishmania genus and is transmitted through sandflies. This disease imposes a significant global health burden, particularly in regions with limited access to healthcare facilities. The parasite undergoes two distinct stages of development: amastigote and promastigote, each playing pivotal roles in the infection process. Various species of sandflies, including Sergentomyia and Phlebotomus, serve as vectors for disease transmission. This study addresses the challenges encountered in drug discovery for leishmaniasis, emphasizing the critical need for effective and safe treatments. Presently available therapeutics exhibit limitations, including adverse side effects and the emergence of drug-resistant strains. Moreover, the pharmaceutical industry's market-driven approach has led to a dearth of innovations for neglected tropical diseases like leishmaniasis. To confront this issue, we propose an innovative approach combining machine learning with cheminformatics to classify drugs as either active or inactive against leishmania promastigote. The study leverages a dataset comprising 65,057 molecules sourced from the PubChem database, employing the Alamar Blue-based assay to assess their susceptibility to various drugs. Molecular fingerprints, derived from Simplified Molecular Input Line Entry System (SMILES) notations, are employed for data encoding. Three distinct types of fingerprints, namely Avalon Fingerprint, MACCS Key Fingerprint, and Pharmacophore Fingerprint, are utilized to train machine learning models. These models aim to accurately categorize molecules according to their characteristics and chemical structure, potentially revolutionizing the approach to drug discovery for leishmaniasis. The study's significance lies in its potential to expedite the drug discovery process, address the global impact of leishmaniasis, and serve as a model for tackling other neglected tropical diseases.
