AI-enhanced 3D-QSAR screening of fragment-based novel designed molecules targeting Phalaris minor ACCase

Abstract

Acetyl-CoA carboxylase (ACCase: EC 6.4.1.2) is a crucial enzyme for fatty acid synthesis in plants, particularly in the Graminae family, making it an ideal target for herbicides aimed at selective weed control in agriculture. One persistent challenge is the infestation of Phalaris minor in wheat (Triticum aestivum) fields, leading to significant crop yield losses. While herbicides are the primary solution to manage P. minor, their overuse has led to resistant biotypes, driving the need for novel herbicide molecules. Leveraging artificial intelligence (AI) and machine learning (ML) in the agritech revolution, researchers are now applying advanced computational techniques to identify and design effective ACCase inhibitors. Using small molecule databases such as ZINC, CHEMBL, and DrugBank, an initial screening based on structural similarity to known ACCase inhibitors is performed. AI-driven high-throughput virtual screening (HTVS) then filters these candidates followed by physiochemical properties based screening. The selected herbicide-like molecules are further processed through fragment-based design to generate a library of new compounds, refined using binding affinity thresholds (-8.5 kcal/mol) and Quantitative Structure-Activity Relationship (QSAR) models. Finally, molecular dynamics (MD) simulations validated the interaction stability of these potential herbicides over 100 ns, yielding four promising candidates optimized for ACCase inhibition. This study showcases how AI-powered methodologies are transforming agricultural science by facilitating the design of next-generation herbicides that can address resistant weed biotypes, underscoring the role of technology in sustainable crop protection.