In-silico prediction of multi‑target mechanisms of Pinellia ternata phytochemicals in lung cancer: Evidence from a graph‑attention‑guided virtual screening and multi‑scale simulations

by Guoqiang Bian, Yuanbin Zhang, Yuanhao Shen, Pengcheng Xiao, Daifeng Zhang, Jiadong Xie, Xiong Li, Duo Chen, Kongfa Hu, Chenjun Hu Pinellia ternate has long been used to treat respiratory diseases, possessing potential anti-tumor activity and exhibiting multi-component, multi-target characteristics. This study prioritized lung cancer-related targets using the HERBGAT framework based on graph attention networks (GAT).

High-quality PDB structures were retrieved, and diffusion-generative docking was performed to construct complex conformations and assess their confidence levels. Molecular dynamics simulations of representative complexes were conducted over 200 ns, and binding free energies were estimated using the MM/PBSA method.

The pharmacokinetic characteristics of the bioactive compounds were evaluated using Swiss ADME and PreADMET computational tools, and density functional theory (DFT) analysis using ORCA software was combined to explore their electronic structure and properties. In this study, the potential targets of Pinellia ternata highly overlap with lung cancer pathological genes, with FGFR4, CDK2, JAK2, KDR, PAK4, PTK2 and PDGFRA being the core.

Baicalein exhibits a conserved binding mode of “hinge hydrogen bond-aromatic interlayer-hydrophobic groove” at targets such as PTK2/KDR/JAK2.