In Silico Drug-Likeness, ADMET, and Toxicity Profiling of Phytochemicals from Annona squamosa

Abstract

Annona squamosa is a rich source of bioactive phytochemicals with documented cytotoxic properties, yet the drug–like potential and safety profiles of its constituents remain systematically unexplored. This study comprehensively characterizes the in silico drug–likeness, pharmacokinetic, and toxicity landscape of a GC–MS–identified phytochemical library from Annona squamosa to prioritize leads for drug discovery. A library of 165 compounds was evaluated using SwissADME and AdmetSAR against Lipinski, Veber, Ghose, and Muegge filters. ADMET properties (absorption, distribution, metabolism, excretion, toxicity) were predicted, and multi–endpoint toxicity profiling was conducted. Principal Component Analysis identified chemical space clustering. Lipinski compliance was high (96.3%), while the more stringent Veber (63.6%), Ghose (31.5%), and Muegge (14.5%) filters caused greater attrition (30.3% passing ≥4 filters). Absorption was predicted to be excellent (97.0% high HIA, 99.4% Caco–2), with 98.2% predicted BBB penetration and 92.1% optimal plasma protein binding. Mitochondrial toxicity was predicted as the most prevalent concern (86.1%), while hERG inhibition (3.0%) and Ames mutagenicity (3.0%) were predicted to be rare. The ADMET funnel retained 15.2% (25 compounds) after all filters. Forty–seven compounds (28.5%) qualified as leads (passing ≥4 drug–likeness filters with ≤2 toxicity alerts), with diterpene alcohols representing the most balanced class. BA–40 emerged as the top candidate (QED 0.715). Annona squamosa phytochemicals are enriched in developable scaffolds. Although mitochondrial toxicity was predicted to be prevalent, genotoxic and cardiotoxic liabilities were predicted to be rare (3.0% each), supporting prioritization of the lead candidates for experimental validation rather than direct therapeutic progression.