Search Results (186)

Background: Colchicine is a microtubule-targeting anti-inflammatory agent with emerging relevance in cardiovascular disease; however, its effects on injury-induced vascular remodeling remain incompletely defined. Methods: In this study, a rat iliac artery clamp injury model was used to evaluate the effects of colchicine (0.5 mg/kg/day, oral gavage) over 28 days. Histomorphometric, histopathological, and immunohistochemical analyses were performed to assess vascular remodeling. In parallel, molecular docking and STRING/Cytoscape-based protein–protein interaction (PPI) network analyses were conducted to provide structural and systems-level context. Results: Colchicine significantly reduced intimal thickness, the intima-to-media (I/M) ratio, luminal stenosis, adventitial thickness, and collagen deposition, while preserving the lumen area and improving the remodeling index. Medial thickness was not significantly affected. Proliferative activity showed a decreasing trend without statistical significance. Circulating inflammatory cytokines, including TNF-α and IL-1β, did not differ significantly between groups. Docking analyses suggested potential interactions with β-tubulin, ADAM17, NLRP3, IKKβ, and RELA, while network analysis identified an interaction architecture centered on NF-κB-related regulatory components and inflammasome-associated signaling pathways. Conclusions: Colchicine attenuates injury-induced vascular remodeling in this experimental model. These findings, together with complementary in silico analyses, suggest a multi-target, inflammation-associated framework involving NF-κB-related and inflammasome-linked pathways. The in silico analyses provide supportive mechanistic context but do not establish causal relationships. Full article

Background: Intervertebral disc degeneration (IVDD) is driven by the interplay between inflammatory signaling, extracellular matrix (ECM) degradation, and impaired cellular adaptation. Although several nutraceutical compounds have been reported to exert protective effects in IVDD-related models, their multitarget mechanisms within integrated molecular networks remain incompletely characterized. Methods: An in silico framework integrating molecular docking with network-based analyses was employed to evaluate resveratrol, quercetin, melatonin, curcumin, and baicalein against a predefined panel of IVDD-associated targets, within an exploratory in silico framework. Binding affinities and interaction profiles were assessed using molecular docking, followed by protein–protein interaction (PPI) network construction, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses, and hub gene identification. Results: Docking analyses revealed binding energies ranging from −4.59 to −13.25 kcal/mol, with curcumin and quercetin showing plausible docking poses across a subset of selected targets under the applied protocol. Network analysis showed a highly interconnected structure centered on key inflammatory regulators, including NFKB1, IL6, TNF, IL1B, STAT3, and NLRP3, together with ECM-associated components such as ACAN, COL2A1, SOX9, MMP13, and ADAMTS5. Enrichment analyses further suggested significant associations with inflammatory signaling pathways, cytokine regulation, and ECM organization. Conclusions: These findings are compatible with a distributed, multitarget interaction pattern of nutraceutical compounds within IVDD-associated molecular networks. By integrating molecular docking with network-based analyses, this study offers a system-level framework for interpreting previously reported effects within a disease-specific context. Docking-derived interaction patterns should be interpreted as qualitative and exploratory observations, as docking scores represent model-dependent estimates and do not establish comparable pharmacological effects across heterogeneous targets. The results should be considered hypothesis-generating and require experimental validation. Full article

Diabetic kidney disease (DKD) affects approximately 40% of patients with diabetes mellitus and remains a leading cause of end-stage renal disease worldwide. Early diagnosis and identification of therapeutic targets are critical for improving patient outcomes, yet reliable biomarkers are lacking. This study integrated transcriptomic data from the Gene Expression Omnibus (GEO) database (GSE96804, GSE30528, and GSE142025) with machine learning algorithms and Mendelian randomization (MR) to identify diagnostic biomarkers for DKD. Differentially expressed genes (DEGs) were identified and intersected with key modules from weighted gene co-expression network analysis (WGCNA). Four machine learning methods—least absolute shrinkage and selection operator (LASSO), random forest (RF), support vector machine-recursive feature elimination (SVM-RFE), and extreme gradient boosting (XGBoost)—were applied for feature selection. Five hub genes (SPP1, CD44, VCAM1, C3, and TIMP1) were identified at the intersection of these approaches. Two-sample MR analysis using eQTL data from the eQTLGen Consortium and kidney function GWAS from the CKDGen Consortium provided evidence supporting potential causal associations between SPP1, C3, and TIMP1 expression and estimated glomerular filtration rate decline. Immune infiltration analysis via CIBERSORT estimated elevated proportions of M1 macrophages and activated CD4${}^{+}$ memory T cells in DKD samples, with all five hub genes showing correlations with macrophage infiltration. A diagnostic model based on these five genes achieved a cross-validated area under the receiver operating characteristic curve (CV-AUC) of 0.938 in the discovery dataset and AUC values of 0.917 and 0.889 in two independent external validation cohorts. Drug–gene interaction analysis identified 10 candidate compounds targeting the hub genes. These findings provide a computational framework for identifying candidate diagnostic biomarkers and generating hypotheses regarding potential therapeutic targets for DKD; however, all results are derived from in silico analyses and require experimental validation—including qPCR, immunohistochemistry, and prospective clinical cohort studies—before clinical applicability can be established. Full article

Background/Objectives: Paracetamol is a widely analgesic and antipyretic drug; however, its limited anti-inflammatory efficacy and safety concerns motivate the search for novel non-opioid alternatives. In this study, a series of 4-hydroxyphenylamino-naphthoquinones were designed as paracetamol-inspired analogs and synthesized via a solvent-free, silica-assisted Michael addition, providing a sustainable and efficient synthetic route. Methods: The compounds were evaluated using an integrated strategy combining in silico prediction, density functional theory calculations, molecular docking, ADMET profiling, and in vivo phenotypic pharmacological assays. Results: In vivo evaluation revealed pronounced peripheral antinociceptive activity in the acetic acid-induced writhing model and robust anti-inflammatory effects in carrageenan-induced paw edema, comparable to those of naproxen. These findings suggest a predominantly peripheral mechanism consistent with anti-inflammatory and antinociceptive profiles linked to cyclooxygenase inhibition. A normalization-based multi-criteria analysis integrating peripheral, anti-inflammatory, central, and antipyretic endpoints enabled transparent phenotypic prioritization within the series. Under this framework, compound 7 emerged as the most balanced peripheral–anti-inflammatory candidate, whereas compound 8, evaluated experimentally as a regioisomeric mixture, showed comparatively stronger central antinociceptive activity in the hot plate test. Antipyretic activity in an LPS-induced fever model was limited and not sustained. Conclusions: Overall, these findings indicated that the 4-hydroxyphenylamino-naphthoquinone scaffold emerges as a promising non-opioid platform for peripheral inflammatory pain, supporting further investigation of its pharmacological and mechanistic properties. Full article

Nepetalactone (NL) is a volatile iridoid monoterpene widely used in biopesticidal and repellent applications, yet its toxicokinetic behavior and metabolic fate as a pure compound remain poorly characterized. This study aimed to provide an integrated toxicokinetic evaluation of NL by combining in silico absorption, distribution, metabolism, excretion and toxicity (ADMET) modeling with in vitro metabolism assays using rat and human liver microsomes, supported by UHPLC–MS/MS analysis for metabolite identification. The in silico biotransformation predicted extensive phase I oxidation followed by phase II conjugation, while ADMET predictions indicated low systemic persistence and limited toxicological concern for most metabolites. The performed in vitro microsomal assays confirmed the in silico prediction by a rapid and time-dependent NL metabolism via both oxidative (86% reduction in NL concentration after 120 min) and conjugative (89% reduction in NL concentration after 120 min) pathways in rat and human systems, with comparable depletion kinetics between species. UHPLC–MS/MS enabled the identification of multiple phase I and phase II metabolites, pointing to pronounced interspecies differences in conjugative metabolism. In this sense, while oxidoreduction and hydrolysis reactions were consistent with previously reported iridoid metabolism. This study suggests the possible formation of previously unreported amino acid-related derivatives, although these require further confirmation. Overall, these findings advance the understanding of NL biotransformation, propose a new, previously unknown, metabolic pathway for iridoids, and provide relevant data to support human health and environmental risk assessment frameworks. Full article

Background/Objectives: Craniofacial malformations such as orofacial clefts affect ~1 in 700 births; 40–60% lack clear genetic etiology, and many exhibit asymmetry and variable expressivity unexplained by classical Sonic Hedgehog (SHH) morphogen gradient models. We investigated whether integrated molecular modules linking morphogen signaling with metabolic stress responses may better account for craniofacial developmental outcomes. Methods: Sequential UniProt gene set integration identified 186 candidate craniofacial regulators. STRING network analysis revealed modular architecture. Molecular docking profiled 17 compounds against SMO, CK1δ, PINK1, and TIE2 (control). Pathway reconstruction integrated the SHH–CK1δ–HIF1A–HEY1–PINK1 axis with in-silico-predicted CK1δ phosphorylation sites on SMO (S615, T593, S751), HIF1A (Ser247), and GLI1/2/3 transcription factors. A developmental decision tree mapped affinity profiles to node-specific phenotype hypotheses. Results: CK1δ and PINK1 emerged as candidate nodes coupling morphogen signaling with mitochondrial quality control. Cross-docking showed preferential binding to developmental kinases (CK1δ: −8.34 kcal/mol; PINK1: −8.80 kcal/mol) versus TIE2 control (−6.76 kcal/mol; p < 0.001). Pathway reconstruction suggested that CK1δ-mediated Ser247 phosphorylation of HIF1A disrupts ARNT dimerization, redirecting HIF1A toward ARNT-independent HEY1 induction and consequent PINK1 suppression. Based on computed profiles, node-specific associations were proposed as computational hypotheses: SMO perturbation → midline defects; CK1δ → facial asymmetry/clefting; PINK1 → mandibular hypoplasia. Multi-target compounds (e.g., purmorphamine, taladegib) generated composite phenotype predictions consistent with clinical complexity. Conclusions: This strictly in silico study identifies candidate integrated morphogenic modules whose multi-node perturbation may underlie anatomically specific craniofacial malformation patterns. Node–phenotype associations are prioritized computational hypotheses requiring experimental validation; if confirmed, the framework could inform developmental toxicity assessment, therapeutic design, and reclassification of idiopathic craniofacial anomalies. Full article

An efficient bacteriophage delivery system needs to be developed to overcome the challenges associated with phage instability, rapid diffusion, and loss of infectivity at the infection site. Hydrogels have been found to be potential carriers. Hydrogels have emerged as promising carriers due to their biocompatibility, tunable physicochemical properties and capacity for controlled release. However, the molecular factors that regulate phage–hydrogel interactions remain poorly understood. In this study, we employed an in silico framework combining molecular docking, molecular dynamics (MD) simulations, MM/PBSA binding energy calculations, machine learning-based adhesion prediction, and diffusion modeling to explore phage–hydrogel interactions at the molecular level. Surface-exposed bacteriophage proteins, such as capsid and tail proteins, were evaluated against eight different hydrogel polymers. Binding site analysis revealed the presence of multiple solvent-accessible pockets that can interact with the polymer. Docking studies showed favorable and stable interactions, with hyaluronic acid showing strong binding affinity to multiple phage proteins (−5.5 to −5.7 kcal/mol) and GelMA showing high affinity to the capsid gp10 protein (−5.6 kcal/mol). The integrity of the structural complexes was further confirmed by 100 ns MD simulations, stable RMSD and RMSF trajectories, compact structural conformations, and favorable MM/PBSA binding energies. Machine learning classification successfully differentiated high- and low-adhesion systems and identified hydrogen bonding and electrostatic interactions as key determinants of sustained yet reversible phage retention. Collectively, our findings suggest that the hydrogels enriched with charged and polar functional groups can facilitate stable but non-destructive phage binding, enabling controlled and sustained release. This study provides mechanistic insights into rational hydrogel design for phage delivery systems and highlights the potential of high-throughput computational strategies to accelerate the development of optimized phage therapeutics. Full article

Optic neuropathy represents a leading cause of irreversible vision loss, in which oxidative stress, chronic inflammation, dysregulated lipid metabolism, and mitochondrial dysfunction contribute to the progressive degeneration of retinal ganglion cells (RGCs). In recent years, a number of nutraceuticals have been investigated as potential neuroprotective agents; however, the molecular mechanisms through which they exert their effects remain incompletely understood and are often considered in isolation. In the present in silico study, an integrative network-based approach was applied for a systematic analysis of the predicted molecular targets of selected nutraceuticals with antioxidant and anti-inflammatory potential. By combining target prediction, protein–protein interaction analysis, and functional enrichment, their functional convergence was assessed in the context of optic nerve pathophysiology. The results indicate that, despite their chemical and functional heterogeneity, the investigated nutraceuticals do not act through fully independent mechanisms but instead converge on interconnected regulatory axes. In particular, lipid–inflammatory signaling, epigenetic and stress-adaptive mechanisms, as well as nuclear-receptor mediated transcriptional regulation emerged as key pathways. These pathways form integrated molecular models potentially determining cellular susceptibility to injury and the adaptive capacity of RGCs. In conclusion, the present analysis provides a systems-level framework for understanding the neuroprotective potential of nutraceuticals, highlighting the importance of network convergence and multi-target activity. The obtained results support the conceptual shift from isolated antioxidant strategies towards integrative, network-oriented approaches in the study of optic neuropathy. Full article

Background: Crimean-Congo hemorrhagic fever (CCHF) is a severe tick-borne viral disease with a high fatality rate, and no licensed vaccines are currently available. The nucleoprotein (NP) of the Crimean-Congo hemorrhagic fever virus (CCHFV) plays a critical role in viral replication and immune recognition, making it a promising target for vaccine development. This study aimed to design and evaluate a multiepitope recombinant DNA vaccine targeting the NP of CCHFV. Methods: Cytotoxic T lymphocyte (CTL) epitopes from the NP were predicted via immunoinformatics approaches and systematically assessed for antigenicity, allergenicity, toxicity, hydrophobicity, and global population coverage. The selected epitopes were incorporated into four DNA vaccine constructs driven by a cytomegalovirus promoter, adjuvanted with human β-defensin 3 (hBD3), and fused to the reporter protein mRuby3. The constructs were evaluated in vitro using a fluorescent reporter system designed to provide a readout of TCR signaling upon the co-culture of T lymphocytes with differentiated monocytic cells expressing antigens. In vivo immunogenicity and protective efficacy were assessed in BALB/c (exploratory pilot) and IFNAR^−/− mice, a highly susceptible model for viral infection. Cytokine responses were measured to assess immunogenicity. Results: In vitro assays showed predominantly antigen-independent T-cell activation, suggesting that nonspecific stimulation inherent to the reporter co-culture system likely obscured the detection of antigen-specific TCR signaling. In vivo analyses in BALB/c mice revealed that the constructs elicited only modest systemic cytokine profiles while CCHFV-specific IgG and IFN-γ secretion remained undetectable, indicating that antigen-specific T-cell and antibody responses were limited. In the IFNAR^−/− challenge model, several peptide groups achieved significant 2–3 log reductions in tissue viral RNA and infectious titers (p < 0.05 vs. sham). However, the observed viral modulations were insufficient to reach the protective threshold and did not translate to a survival benefit (0%). Conclusion: Despite a rational in silico foundation, the multiepitope DNA vaccine constructs demonstrated limitations in inducing potent, antigen-specific immunity across both mouse models. The lack of antigen-specific responses indicates limitations in epitope selection, construct design, and delivery strategies, requiring optimization of next-generation epitope-based vaccines. These findings highlight the complexity of translating computational epitope predictions into functional vaccines, and provide benchmark data as a framework to guide future optimizations. Full article

Background: Breast cancer (BC) is one of the most diagnosed malignancies and a leading cause of cancer-related mortality among women worldwide, thereby posing a substantial threat to women’s health worldwide. However, clinically robust diagnostic biomarkers with high sensitivity and specificity, as well as well-validated molecular targets for targeted therapy, remain limited. Methods: BC transcriptomic data from seven GEO datasets and the TCGA-BRCA cohort (n = 1231) were integrated for analysis. After batch-effect correction, candidate genes were screened through DEA, WGCNA, and PPI networks analysis. An ensemble machine learning (ML) framework incorporating 127 algorithmic combinations was constructed, and SHAP analysis was applied to identify hub genes. Further analyses included functional enrichment, immune infiltration, miRNA regulatory network analysis, and SMR analysis. The expression patterns were validated using single-cell transcriptome data. Drug repositioning analysis and AI-assisted virtual screening were performed to prioritize compounds with favorable drug-like properties. The predicted binding modes of candidate compounds with CHEK1 were assessed by molecular docking. Results: Thirty core genes were obtained through differential expression, WGCNA, and PPI screening. Integrated ML (127 algorithms) determined the optimal model (AUC = 0.919), and SHAP identified nine feature genes, among which CHEK1 and KIF23 showed preliminary diagnostic potential across four external cohorts (AUC: 0.625–0.938). Functional enrichment indicated that both are enriched in the cell cycle and p53 pathways, closely associated with BRCA1/ATR; immune infiltration revealed significant correlations with macrophages and CD8⁺ T cells, with hsa-miR-15a-5p and hsa-miR-607 being common upstream regulatory miRNAs. SMR analysis supported a causal relationship between CHEK1 expression and BC genetic susceptibility (p_SMR < 0.05, p_HEIDI > 0.05); single-cell analysis confirms its heterogeneous expression. AI-assisted virtual screening identified 25 A-grade computational candidate compounds from 171 candidates. Molecular docking suggested that Olaparib and LY294002 can form favorable interactions with the CHEK1 active pocket. Conclusions: The study identified CHEK1 as a key diagnostic gene for BC through 127 ML algorithms and SMR causal inference. By combining AI-assisted virtual screening and molecular docking, computational candidate compounds targeting CHEK1 were prioritized. These findings represent hypothesis-generating in silico predictions and require experimental validation before any therapeutic conclusions can be drawn. Full article

In response to recent advances in computer-aided drug discovery (CADD) enabled by high-performance computing, computational approaches were employed to support and rationalize the investigation of a VEGFR-2-targeted anticancer candidate, combining molecular-level modeling with experimental validation. Initial in silico ADMET profiling and molecular docking were conducted to support the evaluation of drug-like properties and target engagement within a series of para-toluidine-based derivatives (1–14). The most biologically active compound was further evaluated through 100 ns molecular dynamics simulations and comprehensive DFT calculations to investigate binding stability and electronic characteristics. Based on a rational design strategy and supported by computational analyses, the compounds were synthesized and fully characterized using IR, MS, ¹H/¹³C NMR, and elemental analysis. Biological evaluation was performed against HepG-2, MCF-7, HCT-116, and normal WI-38 cells. Mechanistic studies included VEGFR-2 inhibition, wound-healing migration assays, cell-cycle distribution analysis, apoptosis assessment, and caspase-3 activation. Several derivatives exhibited micromolar cytotoxic activity, with compound 14 emerging as the most active against HepG-2 cells (IC₅₀ = 7.84 ± 0.5 µM), showing cytotoxic activity comparable to that of sorafenib (IC₅₀ = 9.18 ± 0.6 µM) and demonstrating favorable selectivity toward normal WI-38 cells (IC₅₀ = 67.75 ± 3.6 µM). Compound 14 showed moderate VEGFR-2 inhibitory activity (IC₅₀ = 0.55 µM), significant suppression of cell migration, pronounced G₀/G₁ cell-cycle arrest, and robust apoptosis induction supported by caspase-3 activation. Molecular docking and MD simulations supported a stable binding mode within the VEGFR-2 active site. This integrated framework highlights compound 14 as a selectively active VEGFR-2-oriented anticancer candidate scaffold with a favorable selectivity profile, supported by experimental and computational analyses, warranting further lead optimization. Full article

Structure-based strategies are widely used in tuberculosis drug discovery; however, their translational impact remains limited. This review examines how structure-based virtual screening (SBVS) is applied in practice to Mycobacterium tuberculosis targets and explores why docking-derived predictions frequently fail to translate into measurable biological activity. Rather than treating docking scores as quantitative predictors of potency, representative case studies are analyzed to demonstrate that SBVS is most effective when employed as a prioritization framework integrated with appropriate target preparation, physicochemical filtering, and early experimental validation. Across diverse targets, molecular dynamics simulations emerge as a critical discriminator, enabling the identification of binding instability and false-positive hits that persist after static docking. Tuberculosis-specific constraints—including cofactor-dependent catalysis, resistance-associated mutations, membrane-rich environments, and permeability barriers—are discussed as key factors decoupling in silico affinity from whole-cell efficacy. Collectively, these observations support a workflow-oriented view of computational drug discovery in tuberculosis, in which iterative integration of structural modeling and experimental validation is required for meaningful lead identification. Full article

Background/Objectives: Protein–protein interactions (PPIs) involving oncogenic drivers remain among the most intractable targets in cancer biology due to their dynamic conformations and limited accessibility to conventional small molecules. Although antibodies and inhibitors have achieved clinical success against targets such as PD-1/PD-L1 and MYC, challenges persist related to tissue penetration, intracellular delivery, resistance, and incomplete blockade of key interface hotspots. The objective of this study is to develop an integrated computational framework for systematically designing hotspot-conditioned de novo miniprotein binders to target these interfaces. Methods: We present DesignForge, a computational protein design pipeline that integrates energetic hotspot identification, generative backbone design, sequence optimization, and structural confidence evaluation. The framework combines hotspot mapping using an open force-field-based energetic analysis module with generative backbone sampling using BindCraft, sequence optimization using ProteinMPNN, and structural validation using AlphaFold2. This in silico pipeline was applied to three representative oncogenic interfaces: PD-1/PD-L1, MYC/MAX, and KRAS/RAF. Results: Computationally generated designs exhibited high predicted structural confidence, favorable interface energetics, and consistent engagement of identified hotspot residues across targets. AlphaFold2-Multimer structural modeling indicated that the candidate PD-1 mimetic scaffolds, MYC/MAX interface binders, and KRAS interaction candidates can adopt conformations compatible with the target interfaces. Energetic contact analysis further supported predicted engagement of key hotspot residues. These findings support the computational feasibility of hotspot-conditioned binder generation using a unified design workflow. Conclusions: DesignForge provides a reproducible computational framework for hotspot-guided de novo protein binder design targeting oncogenic protein–protein interfaces. The designs reported here represent computational predictions derived from structural modeling and energetic analysis. Experimental biochemical and cellular validation will be required to determine the functional activity of the proposed binders. Full article

Background/Objectives: Serotonin transporter (SERT) inhibition represents a central pharmacological strategy in the treatment of major depressive disorder. In this study, an integrated computational framework combining quantitative structure–activity relationship (QSAR) modeling, molecular docking analysis, and in silico ADMET profiling was applied to identify and prioritize novel candidate structures. Methods: Conformation-independent QSAR models were developed using local molecular graph invariants and SMILES-based descriptors optimized through a Monte Carlo learning procedure, while a genetic algorithm–multiple linear regression (GA–MLR) was employed to derive statistically robust predictive models from a large descriptor pool. Model quality, robustness, and external predictivity were rigorously evaluated using multiple statistical validation criteria. In parallel, a field-based contribution analysis was applied to construct a three-dimensional QSAR model, enabling spatial interpretation of structure–activity relationships. Fragment-level contributions associated with activity enhancement or attenuation were subsequently identified and used to design new candidate inhibitor structures. Results: The designed compounds were further evaluated by molecular docking, InducedFit Docking and Binding Pose MetaDynamics (BPMD) into the SERT binding site, providing a structure-based assessment consistent with the trends observed in QSAR modeling. In addition, in silico ADMET analysis was performed to assess key pharmacokinetic and safety-related properties relevant to central nervous system drug development. Conclusions: The proposed workflow demonstrates the utility of combining data-driven QSAR modeling with structure-based and pharmacokinetic considerations to rationalize and prioritize novel serotonin transporter-focused scaffold optimization, offering a transferable strategy for early-stage antidepressant drug discovery. Full article

Endocrine-Disrupting Chemicals (EDCs) disrupt thyroid hormone (TH) synthesis, transport, metabolism, and action, thereby perturbing systemic endocrine homeostasis. Transthyretin (TTR) is a key TH transport protein that regulates circulating hormone distribution and tissue availability, particularly during critical developmental windows. Chemical interference with TTR-binding may alter TH bioavailability and represent a transport-mediated molecular initiating event within thyroid-axis perturbation. Despite widespread exposure, many thyroidal EDCs remain unidentified, and their health effects are difficult to assess due to multiple simultaneous exposures. To support endocrine hazard identification and chemical prioritization within risk assessment frameworks, we developed machine learning-based QSAR models during the Tox24 challenge, using a dataset of 1512 chemicals to predict TTR-binding affinity. Of these, 67% were used for training, 13% for testing, and 20% for validation. Molecular descriptors were selected by first removing highly correlated features and then ranking the remaining descriptors using mutual information regression. The leverage approach was applied to define the models’ applicability domain (AD). Five machine learning algorithms, including gradient boosting regressor (GBR), Random Forest, Lasso Regression, Support Vector Machine (SVM), and regularized SVM models, were developed. The GBR model demonstrated the best overall performance. This model achieved an R² of 0.89 on the training set, 0.58 on the test set, and 0.55 on the validation set. The molecular descriptor analysis highlights hydrophobicity, steric effects, branching, connectivity, and ionization/electronic effects as the mechanistic basis for TTR disruption and stabilization, providing structural insight into features associated with thyroid hormone displacement. The AD analysis indicated that 97.5% of the test set and 96.0% of the validation set fell within the reliable descriptor space. Importantly, these predictions extend beyond model benchmarking by informing weight-of-evidence evaluations of thyroid-axis perturbation and supporting the prioritization of chemicals for targeted testing within non-animal new approach methodologies. Overall, this work highlights the application of in silico approaches for screening EDCs, supporting the prioritization and identification of potentially harmful chemicals. Full article