A Multi-Criteria Computational Decision Framework for Prioritizing Phytochemicals Targeting HSP90AA1: Molecular Docking, MM–GBSA, QikProp and ADME-Based Assessment for targeting Frostbite

Frostbite is a cold injury of the tissues characterized by cellular injury, vascular dysfunction, inflammation and reperfusion injury. Although not yet determined if it has a specific function in the pathophysiology of frostbite, a stress responsive molecular chaperone, encoded by HSP90AA1 (heat shock protein 90 alpha), is a candidate framework for exploratory computational screening. In this study, 14 phytochemicals were molecularly docked and Prime MM–GBSA calculations and QikProp ADME descriptors were obtained to compare the phytochemicals for predicted interaction with the N-terminal domain of HSP90α. Barbaloin gave the best Glide docking score (−7.421) followed by epigallocatechin gallate (EGCG; −7.259), rutin (−6.982) and resveratrol (−6.941). The primary MM–GBSA estimates of EGCG (−51.83 kcal mol−1), rutin (−51.49 kcal mol−1), resveratrol (−46.48 kcal mol−1) and barbaloin (−46.24 kcal mol−1) were most favourably ranked. These discrepancies between docking and MM–GBSA rankings suggest that it is important to use both scores together and not just docking scores alone. The short-listed ligands exhibited different property profiles as predicted by QikProp: Resveratrol was the most highly predicted human oral absorption percentage of these four ligands (82.32%) while EGCG and rutin were low, and the Caco-2 permeability of EGCG and rutin were also low. The contact patterns predicted by the 2D interaction maps within the HSP90α N terminal binding region were different. Based on these results, EGCG, rutin, barbaloin and resveratrol are suggested as the substances to be further studied based on the preliminary computational approach. Calculations do not validate the HSP90AA1 inhibition or therapeutic efficacy against frostbite, biochemical and cold-injury model validations are still needed. From a computational decision-making perspective, the proposed multi-criteria framework integrates molecular interaction scores and predicted pharmacokinetic properties to support transparent phytochemical prioritization. The findings demonstrate how complementary computational indicators can inform evidence-based candidate selection for further experimental investigation while acknowledging the limitations of in-silico predictions.