Molecular detection of blaTEM in multidrug-resistant E. coliand deep learning-assisted computational identification of phytochemical inhibitors from Detarium microcarpum

• Molecular detection confirmed the presence of the blaTEM gene in MDR E. coli. • GC-MS identified 23 bioactive compounds from Detarium microcarpum stem bark. • Deep learning predicted IC₅₀ values and prioritised candidate inhibitors. • Docking and MD simulations validated the strong binding of lead phytochemicals. • CID-538133 emerged as the most stable non-classical β-lactamase inhibitor candidate. The rising prevalence of extended-spectrum β-lactamase (ESBL)-producing Escherichia coli ( E. coli ), particularly those harbouring the blaTEM gene, poses a serious public health challenge. This study aimed to characterise multidrug-resistant (MDR) E. coli isolates from clinical samples and identify potential inhibitors of the blaTEM enzyme from Detarium microcarpum stem bark using deep learning–assisted in silico approaches. PCR amplification and sequencing confirmed the presence of the blaTEM gene (GenBank Accession: OR672059.1). Phytochemical screening of the ethanolic extract revealed flavonoids, alkaloids, saponins, and coumarins, while GC-MS identified 23 compounds. A deep learning regression model trained on molecular descriptors predicted IC₅₀ values, prioritising candidates with potent theoretical inhibition. CID-5363192 (7-hydroxycoumarin derivative) and CID-9601770 showed predicted IC₅₀ values below 1 µM, with favourable physicochemical profiles. ADMET screening selected eight compounds with optimal pharmacokinetics and drug-likeness. Molecular docking revealed strong binding affinities for CID-538133 (-6.27 kcal/mol), CID-543346 (-6.05 kcal/mol), and CID-5363692 (-5.20 kcal/mol), surpassing Avibactam (-5.60 kcal/mol). MD simulations over 300 ns confirmed CID-538133 as the most stable candidate, showing RMSD fluctuations between 2.0-3.0 Å and compactness around 19.5 Å. These results suggest that phytochemicals from D. microcarpum may act as non-classical β-lactamase inhibitors, with CID-538133 emerging as the most promising lead. These findings are hypothesis-generating and provide a computational rationale for future biochemical β-lactamase inhibition assays to validate the predicted inhibitory activity experimentally.