In Silico ADME–Toxicity Profiling of 3,4-dithia-1-azabicyclo[3.2.0]heptan-7-one: Integrating Drug-Likeness, Pharmacokinetics, and Prediction Confidence
DOI:
https://doi.org/10.51317/jmhs.v5i1.1194Keywords:
Acute oral toxicity, applicability domain, computational toxicology, disulphide, drug-likenessAbstract
This study reports a confidence-aware, single-compound in silico ADME–toxicity profile of 3,4-dithia-1-azabicyclo[3.2.0]heptan-7-one (C4H5NOS2; molecular weight 147.22 g/mol), a compact bicyclic scaffold comprising a β-lactam and ring-embedded disulfide. The compound was evaluated from supplied SwissADME and ProTox 3.0 reports, with emphasis on cross-endpoint consistency and the evidential weight of confidence-related indicators. The physicochemical profile was favourable, with moderate polarity, low-to-mild calculated lipophilicity and consistently favourable predicted aqueous solubility. High gastrointestinal absorption, absence of predicted P-glycoprotein substrate behaviour and negative predictions for CYP1A2, CYP2C19, CYP2C9, CYP2D6 and CYP3A4 inhibition supported a favourable preliminary ADME profile. The compound passed Lipinski, Veber and Egan filters but failed Ghose, Muegge and lead-likeness criteria largely because of its unusually small size; a disulfide-associated structural alert identified chemical reactivity as a specific uncertainty. ProTox 3.0 estimated an acute oral LD50 of 500 mg/kg (class 4). Still, the reported prediction accuracy was only 23 per cent and average similarity to reference compounds was 31.74 per cent, materially limiting confidence in the point estimate. Neurotoxicity, respiratory toxicity, BBB-barrier activity, ecotoxicity and clinical toxicity were flagged, whereas several other organ and pathway endpoints were inactive. The integrated profile therefore supports conditional compound prioritisation for experimental validation rather than a definitive safety or development conclusion. The principal contribution is a confidence-aware evidence framework that converts computational agreement and discordance into explicit experimental decision questions; no molecular analogue generation, optimisation, docking or SAR modelling was performed.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Edwin Otieno Akumu

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.