Identification of potent phytochemicals against Magnaporthe oryzae through machine learning aided-virtual screening...

Magnaporthe oryzae stands as a notorious fungal pathogen responsible for causing devastating blast disease in cereals, leading to substantial reductions in grain production. Despite the usage of chemical fungicides to combat the pathogen, their effectiveness remains limited in controlling blast disease. Consequently, there exists a pressing need to discover a novel natural biofungicide for efficient blast disease management. To address this challenge, we combined machine learning-based bioactivity prediction with virtual screening, molecular docking, and molecular dynamics (MD) simulations to explore the molecular interactions between forty-eight plant-derived natural compounds and the effector protein, Avr-PikE, an avirulence protein from Magnaporthe oryzae. Among the evaluated phytochemicals, Calotropin, Lupeol, and Azadirachtin emerged as the top-ranking molecules based on their favourable affinity through molecular docking with the effector. MD simulations for 100 ns were conducted to ascertain the stability and reliability of these compounds. Through classical and steered MD simulations and free energy calculations, it was revealed that these selected compounds exhibit stable and favourable energies, thereby establishing strong binding interactions with Avr-PikE. These screened natural metabolites were also found to meet crucial criteria for fungicide-likeness. To support accessibility and broader applications, we also developed a bioactivity prediction app (http://login1.cabgrid.res.in:5260/), allowing users to predict bioactivity against fungi based on our model. The efficacy of one potent compound, Lupeol, was validated through in vitro experiments, confirming its significant antifungal activity against Magnaporthe oryzae. Such biofungicides hold promise for enhancing disease management strategies and mitigating the impact of blast disease on cereal crops.