Potent multi-target natural inhibitors against SARS-CoV-2 from medicinal plants of the Himalaya: a discovery from hybrid machine learning, chemoinformatics, and simulation assisted screening

Identification of novel fructose 1,6-bisphosphate aldolase inhibitors against tuberculosis: QSAR, molecular docking, and molecular dynamics simulation-based analysis of DrugBank compounds

Global initiatives aim to curb tuberculosis (TB) by developing efficient vaccines and drugs against Mycobacterium tuberculosis (M. tb). The pressing need for innovative and swift anti-TB drug screening methods, due to the drawbacks of traditional approaches, is met by employing Structure-based virtual screening (SBVS) and machine learning (ML) in drug discovery. The present study utilizes these methods to repurpose compounds from the DrugBank database (DBD) as anti-TB drugs, explicitly targeting the enzyme fructose-1,6-bisphosphate aldolase (FBA) in glycolysis and gluconeogenesis pathways.Five classifiers, including REPTree, Decision Stump, Random Tree, Random Forest, and J48evaluate training data against M. tbFBA. AdmetSAR 2.0 assesses drug-like properties and toxicity of ML-identified compounds using four filters. Out of 9213 DBD compounds, 5280 were predicted as TB-active. REPTree, chosen for further screening, led to the identification of four promising preclinical anti-TB drug candidates from DrugBank-Serdemetan, Parecoxib, N, N-Diethyl-2-[(2-Thienylcarbonyl) amino], and Visnadine.All screened ligands show stable binding behaviour during a 200-ns molecular dynamics simulation. Density functional theory (DFT) analysis was also employed for the analysis HOMO (highest occupied molecular orbital)/LUMO (lowest unoccupied molecular orbital) gap, and both screened hits showed efficient results. This study presents a potential avenue for effective TB therapeutics development from compounds with proven druggability in other contexts.

Identification of Novel PI3Kα Inhibitor Against Gastric Cancer: QSAR-, Molecular Docking-, and Molecular Dynamics Simulation-Based Analysis

Gastric cancer (GC) is a malignant tumor with global incidence and death ranking fifth and fourth, respectively. GC patients nevertheless have a poor prognosis despite the effectiveness of more advanced chemotherapy and surgical treatment options. The second most frequently mutated gene in GC is PI3Kalpha, a confirmed oncogene that results in abnormal PI3K/AKT/mTOR signaling, causing enhanced translation, proliferation, and survival, and is mutated in 7-25% of GC patients. The protein PI3Kalpha was targeted in the present study by utilizing machine learning (ML), molecular docking, and simulation. A total of 9214 molecules from the DrugBank database were chosen for the first screening. A training set for 6770 compounds tested against PI3Kalpha was assessed to create a quantitative structure-activity relationship-based machine learning model using five different classification algorithms: random forest, random tree, J48 pruned tree, decision stump, and REPTree. Furthermore, consideration was given to the random forest classifier for screening based on its performance index (Kappa statistics, ROC, and MCC). Overall, 1539 of the 9214 drug bank compounds were predicted to be active. Thereafter, three pharmacological filters, Lipinski's rule, Ghose filter, and Veber rule, were applied to test the drug-like properties of the screened compounds. Twenty-six of 1593 compounds showed excellent drug-like properties and were further considered for molecular docking. Thereafter, two compounds were screened as hits because they possessed the molecular docked position with the lowest binding energy and an excellent bonding profile. The binding stability of the selected compounds was further assessed through molecular dynamics simulations for up to 100 ns. Furthermore, compound 1-(3-(2,4-dimethylthiazol-5-YL)-4-oxo-2,4-dihydroindeno[1,2-C]pyrazol-5-YL)-3-(4-methylpiperazin-1-YL) urea was selected as a potential hit in the final screening by analyzing a number of parameters, including the Rg, RMSD, RMSF, H bonding, and SASA profile. Therefore, we conclude that compound 1-(3-(2, 4-dimethylthiazol-5-YL)-4-oxo-2,4-dihydroindeno[1,2-C]pyrazol-5-YL)-3-(4-methylpiperazin-1-YL) urea has efficient inhibitory potential against PI3Kalpha protein and could be utilized for the development of effective drugs against GC.

Identification of small molecule inhibitors against Lin28/let-7 to suppress tumor progression and its alleviation role in LIN28-dependent glucose metabolism

The reciprocal suppression of an RNA-binding protein LIN28 (human abnormal cell lineage 28) and miRNA Let-7 (Lethal 7) is considered to have a prime role in hepatocellular carcinoma (HCC). Though targeting this inhibition interaction is effective for therapeutics, it causes other unfavorable effects on glucose metabolism and increased insulin resistance. Hence, this study aims to identify small molecules targeting Lin28/let-7 interaction along with additional potency to improve insulin sensitivity. Of 22,14,996 small molecules screened by high throughput virtual screening, 6 molecules, namely 41354, 1558, 12437, 23837, 15710, and 8319 were able to block the LIN28 interaction with let-7 and increase the insulin sensitivity via interacting with PPARγ (peroxisome proliferator-activated receptors γ). MM-GBSA (Molecular Mechanics-Generalized Born Surface Area) analysis is used to re-score the binding affinity of docked complexes. Upon further analysis, it is also seen that these molecules have superior ADME (Absorption, Distribution, Metabolism, and Excretion) properties and form stable complexes with the targets for a significant period in a biologically simulated environment (Molecular Dynamics simulation) for 100 ns. From our results, we hypothesize that these identified 6 small molecules can be potential candidates for HCC treatment and the glucose metabolic disorder caused by the HCC treatment.