Machine Learning-Driven Consensus Modeling for Activity Ranking and Chemical Landscape Analysis of HIV-1 Inhibitors DOI Creative Commons

Danishuddin,

Md Azizul Haque, Geet Madhukar

et al.

Pharmaceuticals, Journal Year: 2025, Volume and Issue: 18(5), P. 714 - 714

Published: May 13, 2025

Background/Objective: This study aimed to develop a predictive model classify and rank highly active compounds that inhibit HIV-1 integrase (IN). Methods: A total of 2271 potential inhibitors were selected from the ChEMBL database. The most relevant molecular descriptors identified using hybrid GA-SVM-RFE approach. Predictive models built Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Support Vector Machines (SVM), Multi-Layer Perceptron (MLP). underwent comprehensive evaluation employing calibration, Y-randomization, Net Gain methodologies. Results: four demonstrated intense achieving an accuracy greater than 0.88 area under curve (AUC) exceeding 0.90. at high probability threshold indicates are both effective selective, ensuring more reliable predictions with confidence. Additionally, we combine multiple individual by majority voting determine final prediction for each compound. Rank Score (weighted sum) serves as confidence indicator consensus prediction, through scores in 2D ECFP4-based models, highlighting models' effectiveness predicting potent inhibitors. Furthermore, cluster analysis significant classes associated vigorous biological activity. Conclusions: Some clusters found be enriched while maintaining moderate scaffold diversity, making them promising candidates exploring unique chemical spaces identifying novel lead compounds. Overall, this provides valuable insights into binders, thereby enhancing models.

Language: Английский

Elevating Performance and Interpretability of In Silico Classifiers for Drug Proarrhythmia Risk Evaluations Using Multi-biomarker Approach with Ranking Algorithm DOI
Ali Ikhsanul Qauli,

Nurul Qashri Mahardika T,

Ulfa Latifa Hanum

et al.

Computer Methods and Programs in Biomedicine, Journal Year: 2025, Volume and Issue: 261, P. 108609 - 108609

Published: Jan. 17, 2025

Language: Английский

Citations

1
Machine learning assisted prediction of disperse dye exhaustion on polylactic acid fiber with interpretable model DOI

Shicheng Liu,

Du Chen,

Fengxuan Zhang

et al.

Dyes and Pigments, Journal Year: 2025, Volume and Issue: unknown, P. 112693 - 112693

Published: Feb. 1, 2025

Language: Английский

Citations

0
Semi-Correlations for the Simulation of Dermal Toxicity DOI Creative Commons
Andrey A. Toropov, Alla P. Toropova, Alessandra Roncaglioni

et al.

Toxics, Journal Year: 2025, Volume and Issue: 13(4), P. 235 - 235

Published: March 23, 2025

The skin is the primary pathway for harmful substances to enter body and a susceptible target organ, making compound-induced acute dermal toxicity significant health risk. In this work, possibility of modelling using so-called semi-correlations studied. Semi-correlations are specific case correlations, where one variable takes only two values. For example, 0 denotes absence activity (e.g., toxicity), 1 presence activity. described computational experiments can be carried out by interested readers freely available software CORAL.

Language: Английский

Citations

0
Graph-Theoretic and Computational Analysis of QSAR Molecular Descriptors for Single Chain Diamond Silicates DOI Creative Commons
Sajeev Erangu Purath Mohankumar, Ponnurengam Malliappan Sivakumar,

S. Priyatharshni

et al.

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: March 28, 2025

Abstract This study presents a comprehensive graph-theoretic and computational analysis of Quantitative Structure-Activity Relationship (QSAR) molecular descriptors for Single Chain Diamond Silicates (CSn), crucial class silicate structures defined by their unique connectivity SiO₄ tetrahedra. Various descriptors, including the Atom Bond Connectivity (ABC) Index, Sum (ABS) Augmented Zagreb Index (AZI), (SZI), Geometric Arithmetic (GAI), (AGI), are examined to assess structural, electronic, thermodynamic properties. Through mathematical formulations modelling, this quantifies complexity, stability, patterns CSn, enhancing predictive capabilities QSAR models. The findings underscore significance in characterising networks, with applications spanning materials science, catalysis, geochemistry.

Language: Английский

Citations

0
Advancements in toxicological risk assessment: integrating Ferguson’s principle, computational models, and drug safety guidelines, a comprehensive framework for improving risk assessment and resource management in toxicology DOI
Saurabh Dilip Bhandare

Toxicology Research, Journal Year: 2025, Volume and Issue: 14(3)

Published: May 2, 2025

Abstract This investigative study examines the transfer of maternal medications into breast milk and their potential impact on breastfeeding infants. Significant factors influencing drug transfer, including physiochemical properties composition, are analysed to corroborate judicious administration in nursing mothers. The investigates, evaluates, interprets drugs such as: H|chlorpromazine (New England Nuclear [NEN]), diazepam Roche, C|diclofenac (Ciba-Geigy, 6.6 mCi/mmol, K-277), diclofenac 0.1317), digoxin (Wellcome, 11725), fluphenazine (Squibb 12240), phenytoin (NEN, 46 Ci/mmol, 2315-061), (Parke-Davis 5419972), pirenzepine (Boehringer-Ingelheim-660206), H|prednisolone (Amersham, 67.4 88), warfarin 30), outlining assessing transferability perils notably presented. Ferguson’s principle was leveraged predict toxicity, specifically for central nervous system depressants, elucidating lethality safety evaluation. On top that, advancements toxicological risk assessment were evaluated, articulated as focusing naloxone programs, predictive modelling, quantitative structure–activity relationship (QSAR) applications, toxicogenomics, ordinary differential equation (ODE) models. comparison between assessments biological monitoring highlights prominence evaluating internal dosages. Progress 3D-QSAR modelling augmented its role forecasting chemical while toxicogenomics application ODE models have contributed research. Hence, shift toward alternate toxicity methodologies driven by ethical concerns, budgetary limits, demand more human-relevant data without sacrificing an animal life, which a concern present scientific investigation; fixed machine algorithms, e.g. random forest, Support Vector Machine (SVM), principle, etc.; omics set correlation through tactile programmed computational heuristics decision science.

Language: Английский

Citations

0
Graph-theoretic and computational analysis of QSAR molecular descriptors for single chain diamond silicates DOI Creative Commons
Sajeev Erangu Purath Mohankumar, Ponnurengam Malliappan Sivakumar,

S. Priyatharshni

et al.

Discover Chemistry., Journal Year: 2025, Volume and Issue: 2(1)

Published: May 8, 2025

Language: Английский

Citations

0
Machine Learning-Driven Consensus Modeling for Activity Ranking and Chemical Landscape Analysis of HIV-1 Inhibitors DOI Creative Commons

Danishuddin,

Md Azizul Haque, Geet Madhukar

et al.

Pharmaceuticals, Journal Year: 2025, Volume and Issue: 18(5), P. 714 - 714

Published: May 13, 2025

Background/Objective: This study aimed to develop a predictive model classify and rank highly active compounds that inhibit HIV-1 integrase (IN). Methods: A total of 2271 potential inhibitors were selected from the ChEMBL database. The most relevant molecular descriptors identified using hybrid GA-SVM-RFE approach. Predictive models built Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Support Vector Machines (SVM), Multi-Layer Perceptron (MLP). underwent comprehensive evaluation employing calibration, Y-randomization, Net Gain methodologies. Results: four demonstrated intense achieving an accuracy greater than 0.88 area under curve (AUC) exceeding 0.90. at high probability threshold indicates are both effective selective, ensuring more reliable predictions with confidence. Additionally, we combine multiple individual by majority voting determine final prediction for each compound. Rank Score (weighted sum) serves as confidence indicator consensus prediction, through scores in 2D ECFP4-based models, highlighting models' effectiveness predicting potent inhibitors. Furthermore, cluster analysis significant classes associated vigorous biological activity. Conclusions: Some clusters found be enriched while maintaining moderate scaffold diversity, making them promising candidates exploring unique chemical spaces identifying novel lead compounds. Overall, this provides valuable insights into binders, thereby enhancing models.

Language: Английский

Citations

0