Employing machine learning for identifying antifungal compounds against Candida albicans

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Mycology

Background:

• The emergence of antifungal resistance necessitates the discovery of novel therapeutic agents.
• Machine learning (ML) offers a powerful approach for accelerating drug discovery by predicting compound activity.

Purpose of the Study:

• To assess the efficacy of ML models in predicting antifungal activity against Candida albicans.
• To develop classification and regression models for identifying novel antifungal compounds.

Main Methods:

• Screening of the eMolecules® library using Random Forest (RF), Support Vector Machine (SVM), and LightGBM algorithms.
• Selection of 17 virtual hits based on ML predictions for subsequent in vitro validation.
• In vitro antifungal assays to determine the activity of selected compounds against Candida albicans.

Main Results:

• Eleven out of 17 selected compounds exhibited antifungal activity against Candida albicans.
• Compounds 1 and 17 demonstrated significant inhibition, with minimum inhibitory concentrations (MICs) of 0.51 µM and 0.071 µM, respectively.
• The RF model showed high efficacy in virtual screening and predicting antifungal potential.

Conclusions:

• The developed ML classification and regression models are effective for identifying new antifungal molecules.
• The RF model is a valuable tool for virtual screening in the search for novel antifungal agents against Candida albicans.
• This study highlights the potential of ML in accelerating the discovery of urgently needed antifungal therapies.