Comprehensive In Silico Investigation of Coumarin Derivatives as Androgen Receptor Antagonists: Ligand‐Based Modeling, MolSHAP‐Driven Design and Molecular Dynamics Validation

ABSTRACT

Coumarin derivatives have a wide range of pharmacological activities, including androgen receptor (AR) antagonistic activity, which might be crucial for prostate cancer treatment. To the best of our knowledge, limited computational SAR analysis has been reported for coumarin derivatives as AR antagonists. Here, a range of ligand‐based in silico modelling techniques were employed to explain structural requirements for increased AR antagonistic activity for a set of coumarin derivatives. Consequently, both classification and regression‐based 2D‐QSAR modelling were performed, along with regression‐based 3D‐QSAR, pharmacophore mapping and R‐group analyses. The latter was done by newly developed machine learning‐based MolSHAP program. The two most promising MolSHAP‐designed ligands were further investigated with structure‐based molecular dynamics simulation analyses. The results of molecular dynamics simulations supported the information gathered from ligand‐based in silico modelling. Prediction with ligand and structure‐based in silico techniques helped proposing two MolSHAP‐designed ligands as promising candidates for AR antagonism that should be experimentally validated for confirming their biological and therapeutic relevance. The current work explores open‐access tools and web‐servers to ensure fast reproducibility and at the same time, reports newly developed open‐access Python program, Open3DQSAR_serialProcessing to assist generation of a series of 3D‐QSAR models simultaneously and to facilitate rapid analyses of their results.