Machine learning assisted methods for the identification of low toxicity inhibitors of Enoyl-Acyl Carrier Protein Reductase (InhA)

Abstract

Tuberculosis (TB) is one of the life-threatening infectious diseases with prehistoric origins and occurs in almost all habitable parts of the world. TB mainly affects the lungs, and its etiological agent is Mycobacterium tuberculosis (Mtb). In 2022, more than 10 million people were infected worldwide, and 1.3 million were children. The current study considered the in-silico and machine learning (ML) approaches to explore the potential anti-TB molecules from the SelleckChem database against Enoyl-Acyl Carrier Protein Reductase (InhA). Initially, the entire database of ∼ 119000 molecules was sorted out through drug-likeness. Further, the molecular docking study was conducted to reduce the chemical space. The standard TB drug molecule's binding energy was considered a threshold, and molecules found with lower affinity were removed for further analyses. Finally, the molecules were checked for the pharmacokinetic and toxicity studies, and compounds found to have acceptable pharmacokinetic parameters and were non-toxic were considered as final promising molecules for InhA. The above approach further evaluated five molecules for ML-based toxicity and synthetic accessibility assessment. Not a single molecule was found toxic and each of them was revealed as easy to synthesise. The complex between InhA and proposed and standard molecules was considered for molecular dynamics simulation. Several statistical parameters showed the stability between InhA and the proposed molecule. The high binding affinity was also found for each of the molecules towards InhA using the MM-GBSA approach. Hence, the above approaches and findings exposed the potentiality of the proposed molecules against InhA.

Keywords: Enoyl-Acyl Carrier Protein Reductase; Machine learning; Molecular docking; Molecular dynamics simulation; Tuberculosis.