Smart Simulations for Scarred Livers: Target-Based Computational Models in Cirrhosis Care

Smart Simulations for Scarred Livers: Target-Based Computational Models in Cirrhosis Care | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Smart Simulations for Scarred Livers: Target-Based Computational Models in Cirrhosis Care Roshni Pushpa Raghavan, Muzammil Kabier, Sunil Kumar, Kirti Theresa Alexander, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7704948/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Recent advances in computational drug discovery, including Quantitative Structure-Activity Relationship (QSAR) modelling, molecular docking and molecular dynamics, have paved the way for identifying novel therapeutic candidates targeting key pathways involved in fibrosis and chronic inflammation. The present research retrieved the data from ChEMBL database to develop the binary classification-based machine learning models for four critical targets: Transforming Growth Factor Beta 1 (TGF-β1), Platelet-Derived Growth Factor (PDGF), Inhibitor of Nuclear Factor Kappa-B Kinase Subunit Beta (IKKB), and Tumor Necrosis Factor Alpha (TNF-α). These targets are implicated in diseases such as liver fibrosis, non-alcoholic steatohepatitis (NASH), and Cirrhosis. We discuss the pharmacotherapeutic relevance of predicted active compounds, their mechanisms of action, and implications for clinical pharmacy practice, including drug monitoring and therapy optimization.The study demonstrates how clinical pharmacists may navigate experimental discoveries with practical therapeutic approaches by using computational predictions to inform drug selection, repurposing, and customized patient care.The four machine learning models, from each selected through comparative analysis between 11 classifiers were rigorously validated with 10x10 K-fold cross validation and achieves MCC_ext ≥ 0.54 and Accuracy_ext ≥ 0.74 indicating robust predictive ability. Target based molecular simulations were conducted to chosen pairs of predicted actives and inactives in each target to further validate the discriminatory potential of the model from a simulation point of view and was found to be consistent with the binding scores and dynamics .Furthermore, these models were incorporated into a single web platform with defined pairwise tanimoto similarity-based applicability domain and hosted in the provided link: https://mutli-model-cirrhosis.streamlit.app/ allowing researchers to utilize the predictive models effectively. Moreover, early identification of potential responders through such computational tools may reduce trial-and-error prescribing, minimize adverse effects, and improve overall treatment outcomes in patients with chronic liver diseases. QSAR Drug Discovery Cirrhosis Molecular Docking Molecular Dynamics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction Cirrhosis is the end-stage of chronic liver disease, marked by extensive fibrosis, loss of liver function, and nodular regeneration. Fibrosis and inflammatory diseases represent major clinical challenges due to their chronic progression, complex pathophysiology, and limited treatment options. 1 These conditions are often characterized by persistent immune activation, tissue remodeling, and extracellular matrix (ECM) accumulation, ultimately leading to organ dysfunction and failure. Traditional therapies typically offer symptomatic relief but fail to halt or reverse disease progression. In recent years, target-based approaches have gained momentum, aiming to intervene at specific molecular pathways implicated in disease pathogenesis 2 . Advances in genomics, proteomics, and bioinformatics have facilitated the identification of key targets such as cytokines, growth factors, and signaling molecules. 3 These targets contribute at multiple levels:TGF-β1 is the master regulator of fibrosis in cirrhosis. Its sustained activation results in collagen buildup and scar tissue formation 4 . PDGF supports this by keeping stellate cells activated, contributing to fibrosis expansion. 5 IKKβ/NF-κB signaling maintains a chronic inflammatory environment, which causes continuous liver injury and HSC activation. TNF-α contributes to hepatocyte death and chronic inflammation, both of which are drivers of progressive liver damage and scarring. Therapies directed at transforming growth factor-beta (TGF-β), interleukins, and immune checkpoint regulators have shown promise in preclinical and early clinical studies 6 . Moreover, small molecules, monoclonal antibodies, and RNA-based therapeutics are being developed to modulate these targets with increased specificity and reduced side effects. The integration of personalized medicine strategies further enhances the potential for targeted interventions by considering patient-specific molecular profiles 4 . Despite these advancements, challenges remain in translating molecular targets into effective, safe, and widely accessible treatments. This research explores the current landscape and future potential of target-based therapies for fibrosis and inflammatory diseases, highlighting both breakthroughs and barriers in the field. It gives clear explanation of the drug inhibitory efficacy across the four key protein targets — TGF-β1, PDGF, IKKβ, and TNF-α how they are linked to inflammation, fibrosis, and specifically cirrhosis 7 . Clinical pharmacists play a crucial role in optimizing pharmacotherapy for Chronic diseases like Cirrhosis which navigates a landscape of emerging targeted therapies, biologics, and small molecules. However, the development of novel drugs is time-consuming and costly, necessitating approaches like artificial intelligence and machine learning which accelerate drug discovery and support the faster mode of clinical decision-making 7 . In silico methods, particularly Quantitative Structure-Activity Relationship (QSAR) modeling and molecular docking, have emerged as powerful tools to predict compound activity against specific biological targets. These computational techniques allow rapid screening of large chemical libraries to identify potential drug candidates with favorable binding profiles. The research focuses on four key molecular targets implicated in fibrosis and inflammation: TGF-β1, PDGF, IKKB, and TNF-α. We integrate machine learning-based activity prediction and docking data to explore the pharmacotherapeutic potential of compounds targeting these pathways. Emphasis is placed on translating these findings into clinical pharmacy practice, highlighting how computational insights can guide drug selection, monitoring, and therapeutic optimization. 2. Materials and methods 2.1. Computational Methodology Binary classification machine learning techniques were used to create the predictive models examined in this review, which evaluated drug inhibitory efficacy across four protein targets like TGF-β1, PDGF, IKKB, and TNF-α linked to inflammation and fibrosis.The datasets were curated from the ChEMBL database, comprising between 700 to 1300 compounds per target, following data cleaning procedures such as imputation, removal of duplicates, and chemical structure validation using RDKit 8 – 9 .A classification threshold of 500 nM (pIC50 = 6.3) was set to binarize compounds into active and inactive classes. Model development was proceeded with PoseidonQ software platform, utilizing Morgan fingerprints for molecular representation 10 different classifiers were benchmarked with 10-fold cross-validation, identifying Random Forest, HistGradientBoosting, and Support Vector Machine (SVM) as top performers for various targets. Models were internally validated using rigorous 10x10 K-fold cross-validation and externally validated with independent datasets to ensure robustness and generalizability. Applicability domains were defined based on Tanimoto similarity distributions to assess prediction confidence 11 – 13 . Complementary molecular docking studies were performed using MzDOCK software to validate predicted compound-target interactions 14 . Docking protocols were calibrated through iterative redocking of co-crystal ligands to select optimal scoring functions, ensuring RMSD values below the accepted 2 Å threshold. Predicted active and inactive compound pairs were docked to their respective protein structures, confirming consistency between computational binding affinities and classification results. Together, these methods provide a strong computational foundation for predicting compound efficacy and informing pharmacotherapeutic decisions. Add the details of the MD simulation with procedure and references. 2.2. ML-Model Dataset, Development and Validatory Protocol A total of Binary Classification based QSAR models were developed, capturing inhibitory relationship in four targets consisting of a pair each from anti-fibrotic (TGF- β1 and PDGF) and anti-inflammatory (IKKB and TNF-α) pathways. Initial IC 50 dataset was extracted from ChEMBL database for each target as specified below: TGF- β1 - CHEMBL4439 (1018 Compounds) PDGF - CHEMBL1913 (1285 Compounds) IKKB - Merged CHEMBL3529 and CHEMBL1991 (1340 Compounds) TNF-α – CHEMBL1825 (706 Compounds) Data imputation, duplicate entry removal and RDKit check were performed to clean the datasets. Random 20% dataset was extracted from the Initial dataset for external validation. The Binary classification threshold was set to 500 nM (pIC50–6.30) for binarization of y label (0: inactive and 1: active). A t-SNE plot was generated for visualization of actives/inactives in chemical space to observe isolated cluster formation. Applicability domain for each target were defined on the basis of pairwise tanimoto similarity distribution with Morgan Fingerprint upon which median is determined to assign the threshold for mean tanimoto similarity between 5 nearest compounds in space corresponding to the predicted compound. Comparative analysis of 11 Classification model to identify the best model for each target dataset on the basis of Mathews Correlation Coefficient (MCC) was performed with K-Fold CV set to 10 iteration and the bar plot was produced for comparative analysis. The best model was used to build the model with default parameters for each target with 10x10 K-Fold Cross validation using Morgan fingerprint generating internal validatory metrics (Accuracy, Sensitivity, Specificity and MCC). Furthermore, to determine the true predictive nature, an external validation was conducted with extracted external dataset. 2.3. Deployment of Multi-Model System into Streamlit Cloud The validated multi model system was deployed into streamlit cloud in the form of web application where user can input single SMILES, multiple SMILES through Excel sheet or sketch molecule with the help of chemical sketcher. The web application predicts whether the compound is active or inactive with respect to applicability domain. The threshold of AD is by default the median of tanimoto similarity distribution but is made to be flexible to impose strict applicability domain. The web application can be utilized in the link: https://mutli-model-cirrhosis.streamlit.app/ 2.4. Molecular Docking Classification of Predicted Acitves/Inactives Predicted single pair of actives and inactives from each target model were docked to respective protein targets to identify whether the model prediction could align with binding scores. For each target, protein structure with co-crystal ligand was extracted from RCSB PDB database which is given below 15 – 18 : TGF- β1–1PY5 (co-crystal inhibitor resname: PY1) PDGF – 6JOL (co-crystal inhibitor resname: STI) IKKB – 4KIK (co-crystal inhibitor resname: KSA) TNF-α – 2AZ5 (co-crystal inhibitor resname: 307) Molecular docking study was conducted with software MzDOCK. The single pair of predicted actives and inactives from each target models were prepared using MMFF94 forcefield in protonation state of pH 7.4 19 . The respective protein of each target was prepared by adding polar hydrogens, assigning kollman charges and removing heteroatoms 20 . The binding site was configured at the site of co-crystal ligand of each protein targets with a buffer-space of 4 Å. The choice of scoring function was determined by 5 iterations of redocking to identify the scoring function which produces lower RMSD between the redocked and native pose 21 – 23 . A box plot representing minimum RMSD deviation in each scoring function for each target was generated. the num_modes and exhaustiveness were set to 9 and 8 respectively and the active\inactive pair were docked with best scoring function. The redocking RMSD and superimposition of redocked and native pose for each target was performed for validatory purpose. The binding affinity of random active inactive pair was tabulated. 2.5 Molecular Dynamic simulation Molecular dynamics (MD) simulations were performed to examine the impact of the solvent environment on the structural stability of the protein–ligand complex using the Desmond v7.2 package (Schrödinger Release 2022-4) 24 . The simulations were done on a Dell Precision 7820 Tower workstation running Ubuntu 22.04.1 LTS (64-bit) and outfitted with an Intel Xeon Silver 4210R processor and an NVIDIA RTX A4000 GPU. For all MD calculations, the OPLS4 force field was used. SPC water molecules were used to solvate the protein–ligand complex, which was placed in the middle of an orthorhombic simulation box with a buffer of 10 Å between the protein and the box's edges. Counterions (Na⁺ and Cl⁻) were added at random to neutralize the simulation system, and the complex's shape was used to modify the periodic boundary box's volume. The dynamic behaviour of the complex was evaluated by running the 100 ns MD simulation under NPT conditions. The conformational stability and domain-specific interactions of the complex were assessed employing all Cα atoms in key structural studies, such as protein–ligand interaction profiling, root mean square deviation (RMSD), and root mean square fluctuation (RMSF) 25 – 26 . 3. Results and Discussion 3.1. t-SNE Plot for Active/Inactive Cluster Visualization An important observation is that actives form isolated clusters and inactives are sparse, which affects the specificity of the model but there is an exception in the case of TNF- α were inactives are forming clusters 3.2. Pairwise Tanimoto Similarity Distribution Plot For all 4 targets, the median is low indicating highly sparse and low similarity heterogenous dataset (Fig. 2). 3.3. Comparative Analysis of 11 Classification Models Random Forrest classifier performed best for TGF-beta and IKKB with MCC of 0.65 ± 0.087 and 0.69 ± 0.051 respectively, HistGradientBoosting classifier for PDGF target with MCC of 0.64 ± 0.060 and SVM classifier for TNF-α with MCC of 0.63 ± 0.11 (Fig. 3 ). 3.4. Internal Validation of 4 Models Internal validation demonstrates that all the 4 models across the 10x10 Folds achieved over 0.6 MCC which indicates good performance and an accuracy of over 80%. TGF- β1 model has low specificity due to sparse inactive (Table 1 ). Table 1 Internal Validation Metrics for Each Target After 10x10 K-Fold CV Target Accuracy Sensitivity Specificity MCC TGF-β1 0.880 ± 0.032 0.965 ± 0.024 0.585 ± 0.123 0.629 ± 0.107 PDGF 0.834 ± 0.038 0.887 ± 0.042 0.748 ± 0.072 0.645 ± 0.078 IKKB 0.843 ± 0.033 0.830 ± 0.052 0.858 ± 0.052 0.688 ± 0.065 TNF-α 0.834 ± 0.046 0.755 ± 0.080 0.880 ± 0.053 0.641 ± 0.101 3.5. External Validation of 4 Models The difference between the internal and external validation is minimal indicating robustness of the models and its ability to predict unseen data (Table 2 ). Table 2 External Validation Metrics with Random 20% External Dataset Target Accuracy_ext Sensitivity_ext Specificity_ext MCC_ext TGF-β1 0.892 0.955 0.688 0.687 PDGF 0.790 0.911 0.631 0.574 IKKB 0.840 0.832 0.849 0.678 TNF-α 0.796 0.723 0.832 0.547 3.4. Iterative Redocking with 6 Scoring function for Determination of Ideal Scorer From the plot, Vina performed better in 6JOL and 2AZ5 and Vinardo performed better in 1PY5 and 4KIK (Fig. 4 ). This scoring function were selected for docking due to the precise local optimization of poses close to global minimum. 3.5. Redocking of Co-Crystal Ligand and Superimposition To validate the docking protocol, re-docking of the co-crystallized ligands into their respective protein targets was performed, and the RMSD values between the re-docked and native poses were calculated. The RMSD values obtained were 0.353 Å for 1PY5, 0.388 Å for 6JOL, 0.728 Å for 4KIK, and 0.538 Å for 2AZ5 (Fig. 5 ). All values were below the acceptable threshold of 2.0 Å, confirming the reliability of the docking procedure. The superimposition of the re-docked and native ligand poses is shown in Fig. 5 , where the cyan-colored ligand represents the re-docked pose and the green-colored ligand represents the native crystallographic pose. The close overlap between the two poses demonstrates the robustness and accuracy of the docking methodology used in this study. 3.6. Binding Affinity of Predicted Actives/Inactive Molecular docking was performed to compare the binding affinities of predicted active and inactive compounds against each target protein. The results are summarized in Table 3 . For 1PY5, the predicted active compound showed a binding affinity of − 10.3 kcal/mol, which was stronger than the inactive compound (–8.4 kcal/mol) and close to the co-crystal ligand (–8.8 kcal/mol). For 6JOL, the active compound exhibited the highest binding affinity (–13.6 kcal/mol), consistent with the co-crystal ligand (–13.6 kcal/mol), whereas the inactive compound showed a significantly weaker and unstable binding score (+ 27.8 kcal/mol), indicating poor binding. In the case of 4KIK, the predicted active ligand had a binding affinity of − 10.9 kcal/mol, outperforming the inactive (–8.6 kcal/mol), but slightly lower than the co-crystal ligand (–13.0 kcal/mol).For 2AZ5, the active compound bound with a score of − 9.4 kcal/mol, which was stronger than the inactive (–7.7 kcal/mol) and identical to the co-crystal ligand (–9.4 kcal/mol). Overall, the docking results showed that the predicted active compounds consistently displayed stronger binding affinities compared to the inactives across all targets, further validating their potential as effective inhibitors. Table 3 Binding Affinity of predicted actives\inactive of each target Targets Binding Affinity (kcal/mol) 1PY5 Actives CC1 = NC(= CC = C1)C2 = NN3CCCC3 = C2C4 = C5C = C(C = CC5 = NC = C4)C(= O)N -10.3 Inactive COCC1c2c(-c3ccncc3)c(-c3cccc(C)n3)nn2CC1(C)C -8.4 co-crystal (redock) -8.8 6JOL Actives CC1 = C(C = C(C = C1)NC(= O)C2 = CC = C(C = C2)CN3CCN(CC3)C)NC4 = NC = CC(= N4)C5 = CN = CC = C5 -13.6 Inactive O = C1N/C = C/c2ccc(c(Br)c2)Oc2cc(cc(Br)c2O)CCNC(= O)/C(= N/O)Cc2ccc(c(Br)c2)Oc2cc(cc(Br)c2O)C/C1 = N\O 27.8 co-crystal (redock) -13.6 4KIK Actives C1COCCC1OC2 = C(C = C(C = C2)C3 = NC(= NC = N3)NC4 = CC = C(C = C4)N5CCN(CC5)C6COC6)C#N -10.9 Inactive Cc1noc(-c2ccc3nc(Nc4cc(CN5CCN(CC(F)F)CC5)ccn4)[nH]c3c2)n1 -8.6 co-crystal (redock) -13 2AZ5 Actives COc1ccc2c(c1F)C(= O)N(C[C@@]1(C#Cc3ccc(-c4nc(-c5cnn(C)c5)ccc4OC(= O)C(C)(C)C)c(F)c3)NC(= O)NC1 = O)C2 -9.4 Inactive CCCn1c(= O)n(CC)c(= O)c2cnc3c(OC)cccc3c21 -7.7 co-crystal (redock) -9.4 3.7. Molecular Dynamics The RMSD values for each target protein in complex with the active and inactive compounds were as follows: 2AZ5: The complex with the active compound exhibited an average RMSD of 3.69 Å (range: 1.36–4.59 Å), while the complex with the inactive compound showed a slightly higher average RMSD of 4.09 Å (range: 1.39–5.40 Å). IKK: The active compound complex had an average RMSD of 4.82 Å (range: 1.74–6.87 Å), whereas the inactive compound complex had an average RMSD of 4.25 Å (range: 1.78–6.23 Å). PDGF: The complex with the active compound maintained an average RMSD of 2.00 Å (range: 1.02–2.54 Å), while the inactive compound complex had a slightly higher RMSD of 2.17 Å (range: 0.94–2.79 Å). TGF: The active compound complex showed an average RMSD of 1.90 Å (range: 1.16–2.52 Å), compared to 2.07 Å (range: 1.15–2.62 Å) for the inactive compound complex. These results suggest that, across all targets, complexes with active compounds generally exhibited slightly lower or more stable RMSD profiles, supporting stronger and more consistent binding interactions compared to their inactive counterparts. In addition to RMSD analysis, root mean square fluctuation (RMSF) was employed to assess the flexibility of specific residues in the protein-ligand complexes. RMSF provides critical insights into residue-level mobility, where lower values suggest stable, well-ordered regions often corresponding to secondary structures such as α-helices and β-sheets while higher values typically reflect flexible regions such as loops and terminal residues. Lower RMSF values at binding site residues indicate stronger ligand-protein interactions and conformational stability. For the 2AZ5 complex, the active compound induced lower fluctuations in key binding site residues such as Val13 (0.73 Å), His15 (0.65 Å), Tyr59 (0.77 Å), and Val123 (0.71 Å). In contrast, the inactive compound complex showed elevated fluctuations for comparable residues, such as Tyr59 (1.093 Å), Gly121 (1.03 Å), And Gly122 (1.57 Å), suggesting reduced binding stability. In the IKK complex, residues interacting with the active compound, including Glu19 (3.48 Å), Thr23 (3.52 Å), and Gly27 (3.65 Å), showed higher flexibility but were consistently involved in ligand interactions. Comparatively, the inactive compound complex exhibited lower RMSF values for these residues (Glu19 at 2.11 Å and Thr23 at 1.93 Å), suggesting a more rigid but potentially less interactive binding conformation. The PDGF complex with the active compound displayed moderate fluctuations in residues such as Ala640 (1.71 Å), Ser643 (1.80 Å), and Glu644 (1.89 Å), indicating dynamic but sustained interactions. In contrast, the inactive compound yielded much lower RMSF values for analogous residues, such as Glu644 (0.75 Å) and Leu_651 (1.02 Å), suggesting reduced mobility and potentially weaker engagement. For TGF, the active compound complex showed higher fluctuations in residues like Gly214 (2.37 Å), His285 (1.5 Å), And Glu284 (1.28 Å), reflecting active engagement with the binding site. The inactive compound displayed significantly reduced fluctuations in these regions, such as His285 (1.11 Å) and Gly286 (0.81 Å), indicating less dynamic interaction with the protein core. These RMSF trends highlight a consistent pattern: complexes with active compounds tend to show moderate residue flexibility around the binding site, reflecting both stability and adaptability, whereas inactive compound complexes display either excessive rigidity or limited movement, possibly leading to suboptimal binding and interaction profiles. These observations corroborate the superior binding characteristics of the active ligands and support their candidacy for further optimization. The RMSD and RMSF plots are given in Fig. 6 and Fig. 7 respectively. A Table summarizing RMSF analysis of active and inactive in each target is given in Table 5 . Table 5 Summary of RMSF Analysis for Active and Inactive Compounds Across Target Proteins. Target Protein Compound Type Key Residues (Selected) RMSF Range (Å) Notable Observations 2AZ5 Active Val13, His15, Tyr59, Val123, Ile155 0.55–1.52 Lower fluctuations at active site; stable secondary structures retained Inactive Tyr59, Gly121, Gly122, Lys11, Leu_157 0.76–2.93 Higher fluctuations in active site residues; potential loss of stable interactions IKK Active Glu19, Thr23, Gly27, Tyr98, Gly168 1.29–5.67 Moderate-to-high fluctuations; sustained interactions at flexible active site Inactive Glu19, Thr23, Tyr98, Asp103, Glu149 0.98–2.56 Lower flexibility but possibly weaker ligand accommodation PDGF Active Ala640, Ser643, Glu644, His650, Val_815 0.61–1.89 Balanced flexibility; engagement of key residues via adaptable side chains Inactive Ser643, Glu644, Leu651, Cys814, Val815 0.43–1.02 More rigid binding site; less conformational adaptability TGF Active Gly214, His285, Glu284, Asp290 0.5–2.37 Dynamic interaction at core residues; high engagement flexibility Inactive His285, Arg294, Gly286, Asp351 0.5–1.31 Relatively stable but lower RMSF flexibility; potentially suboptimal binding dynamics 4. Discussion An interesting observation from the internal validation metrics of the models given in Table 1 is MCC of all the target models is above 0.6 indicating good predictive nature, considering a rigorous validation of 10x10 K-fold CV. High sensitivity is seen in TGF-β1 due to isolated active clusters, but poor specificity due to sparse inactives in chemical space. The specificity in TNF-α is particularly high due to clustered inactives. Although, all target models had MCC of above 0.6, TGF-β1 and TNF-α shows higher standard deviation due to its sparse active\inactive data distribution nature. External validation shows that all the target models have good predictive capability considering the dataset was highly heterogenous with natural class distribution imbalances. MCC_ext and ACC_ext of above 0.54 and 0.79 respectively as given in Table 2 within AD definition indicates the model has good predictive power in unseen data. The deviation in MCC between internal and external metric is minimal which establish the robustness of the 4 QSAR models. For further validation, molecular docking studies were conducted with proper redocking protocol beforehand to calibrate the conditions to produce high resolution poses. RMSD of redocking protocol was observed to be below the common threshold of < 2 Å which indicates the scoring function is well adapted to find global minimum. The docking was conducted in the same seed of redocking procedure and the results are given in results section Table 3 . From the observation of Table 4, the binding score of the random pairs of predicted actives and inactives in each target were successfully discriminated by the chosen scoring function, aligning with the model prediction results. This provides evidence of model’s predictive expansion within biological context. Furthermore, the MD simulations provided a much more effective evidence that the prediction of the model could be even be translated in to the compounds effect in the kinetics of the protein. Conclusion The application of QSAR, molecular docking and dynamics targeting TGF-β1, PDGF, IKKB, and TNF-α underscores the potential of computational methods to revolutionize pharmacotherapy in fibrosis and inflammatory diseases. These approaches identify promising therapeutic candidates and provide a robust framework for clinical pharmacists to enhance drug selection, monitoring, and patient care. They bridge the gap between drug discovery and clinical application, ensuring that novel compounds targeting fibrogenic pathways are evaluated for safety, efficacy, and drug interactions. Furthermore, their expertise in pharmacokinetics and pharmacodynamics enables the optimization of personalized therapies for cirrhosis patients, ultimately improving treatment outcomes The provided validated multi-model system with defined applicability domain within a single web-based platform can be of great utility to researchers. The models are reliable and rigorously validated and hence can be used in research application by the scientific community from the given link: https://mutli-model-cirrhosis.streamlit.app/ Declarations Funding The Deanship of Graduate Studies and Scientific Research at Jouf University under grant No. (DGSSR-2025-01-01639) Author Contribution R.P.R. contributed to manuscript writing. S.S. manuscript writing. K.T.A. performed clinical writing and manuscript writing. M. K. and S.K. carried out software development, molecular simulations, computational analysis, and manuscript writing. S.K. and B.M. contributed to conceptualization and supervision. B.M. also provided medicinal chemistry insights, validation, editing, and drafting. M.A.A. and S.N.A.B. were involved in overall supervision and project administration. All authors reviewed and approved the final manuscript. Acknowledgement This work was funded by the Deanship of Graduate Studies and Scientific Research at Jouf University under grant No. (DGSSR-2025-01-01639) References Schuppan D, Afdhal NH (2008) Liver cirrhosis. Lancet 371(9615):838–851. 10.1016/S0140-6736(08)60383-9 Hopkins AL (2008) Network pharmacology: the next paradigm in drug discovery. Nat Chem Biol 4(11):682–690. 10.1038/nchembio.118 Rifai N, Gillette MA, Carr SA (2006) Protein biomarker discovery and validation: the long and uncertain path to clinical utility. Nat Biotechnol 24(8):971–983. 10.1038/nbt1235 Meng XM, Nikolic-Paterson DJ, Lan HY (2016) TGF-β: the master regulator of fibrosis. Nat Rev Nephrol 12(6):325–338. 10.1038/nrneph.2016.48 Borkham-Kamphorst E, Kovalenko E, van Roeyen C-R, Gassler N, Bomble M, Ostendorf T, Floege J, Gressner AM, Weiskirchen R (2008) Platelet-derived growth factor isoform expression in carbon tetrachloride-induced chronic liver injury. Lab Invest 88(10):1090–1100. 10.1038/labinvest.2008.71 Wirth T, Adler G et al (2012) Hepatic activation of IKK/NF-κB signaling induces liver fibrosis via macrophage-mediated chronic inflammation. Hepatology 55(3):1037–1045. 10.1002/hep.24721 Bradham CA, Plümpe J, Manns MP, Brenner DA, Trautwein C (1998) I. TNF-induced liver injury. Am J Physiol Gastrointest Liver Physiol 275(3):G387–G392. 10.1152/ajpgi.1998.275.3.G387 Mendez D, Gaulton A, Bento AP, Chambers J, De Veij M, Félix E, Magariños MP, Mosquera JF, Mutowo P, Nowotka M, Gordillo-Marañón M, Hunter F, Junco L, Mugumbate G, Rodriguez-Lopez M, Atkinson F, Bosc N, Radoux CJ, Segura-Cabrera A, Hersey A, Leach AR (2019) ChEMBL: towards direct deposition of bioassay data. Nucleic Acids Res 47(D1):D930–D940. 10.1093/nar/gky1075 RDKit Open-source cheminformatics. https://www.rdkit.org Kabier M, Gambacorta N, Ciriaco F, Mastrolorito F, Kumar S, Mathew B, Nicolotti O (2025) PoseidonQ: A Free Machine Learning Platform for the Development, Analysis, and Validation of Efficient and Portable QSAR Models for Drug Discovery. J Chem Inf Model 65(8):3944–3954. 10.1021/acs.jcim.4c02372 Bajusz D, Rácz A, Héberger K (2015) Why Is Tanimoto Index an Appropriate Choice for Fingerprint-Based Similarity Calculations? J Cheminform 7(1):20. https://doi.org/10.1186/s13321-015-0069-3 Ciriaco F, Gambacorta N, Alberga D, Nicolotti O (2021) Quantitative Polypharmacology Profiling Based on a Multifingerprint Similarity Predictive Approach. J Chem Inf Model 61(10):4868–4876. https://doi.org/10.1021/acs.jcim.1c00498 Ciriaco F, Gambacorta N, Trisciuzzi D, Nicolotti OPLATO (2022) A Predictive Drug Discovery Web Platform for Efficient Target Fishing and Bioactivity Profiling of Small Molecules. Int J Mol Sci 23(9):5245. https://doi.org/10.3390/ijms23095245 Kabier M, Gambacorta N, Trisciuzzi D, Kumar S, Nicolotti O, Mathew B (2024) MzDOCK: A free ready-to-use GUI-based pipeline for molecular docking simulations. J Comput Chem 45(23):1980–1986. 10.1002/jcc.27390 Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN, Bourne PE (2000) The Protein Data Bank. Nucleic Acids Res 28(1):235–242. 10.1093/nar/28.1.235 He MM, Smith AS, Oslob JD, Flanagan WM, Braisted AC, Whitty A, Cancilla MT, Wang J, Lugovskoy AA, Yoburn JC, Fung AD, Farrington G, Eldredge JK, Day ES, Cruz LA, Cachero TG, Miller SK, Friedman JE, Choong IC, Cunningham BC (2005) Small-molecule inhibition of TNF-alpha. Science 310(5750):1022–1025. 10.1126/science.1116304 Liu S, Misquitta YR, Olland A, Johnson MA, Kelleher KS, Kriz R, Lin LL, Stahl M, Mosyak L (2013) Crystal structure of a human IκB kinase β asymmetric dimer. J Biol Chem 288(31):22758–22767. 10.1074/jbc.M113.482596 Sawyer JS, Beight DW, Britt KS, Anderson BD, Campbell RM, Goodson T Jr, Herron DK, Li HY, McMillen WT, Mort N, Parsons S, Smith EC, Wagner JR, Yan L, Zhang F, Yingling JM (2004) Synthesis and activity of new aryl- and heteroaryl-substituted 5,6-dihydro-4H-pyrrolo[1,2-b]pyrazole inhibitors of the transforming growth factor-beta type I receptor kinase domain. Bioorg Med Chem Lett 14(13):3581–3584. 10.1016/j.bmcl.2004.04.007 Halgren TA, MMFF VII (1999) Characterization of MMFF94, MMFF94s, and other widely available force fields for conformational energies and for intermolecular-interaction energies and geometries. J Comput Chem 20(7):730–748. 10.1002/(SICI)1096-987X(199905)20:7%3C730::AID-JCC8%3E3.0.CO;2-T Wang J, Wolf RM, Caldwell JW, Kollman PA, Case DA (2004) Development and testing of a general amber force field. J Comput Chem. ;25(9):1157-74. doi: 10.1002/jcc.20035. Erratum in: J Comput Chem. 2005;26(1):114 Koes DR, Baumgartner MP, Camacho CJ (2013) Lessons learned in empirical scoring with smina from the CSAR 2011 benchmarking exercise. J Chem Inf Model 53(8):1893–1904. 10.1021/ci300604z Eberhardt J, Santos-Martins D, Tillack AF, Forli S, AutoDock (2021) Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model 61(8):3891–3898. 10.1021/acs.jcim.1c00203 Quiroga R, Villarreal MA (2016) Vinardo: A Scoring Function Based on Autodock Vina Improves Scoring, Docking, and Virtual Screening. PLoS ONE 11(5):e0155183. 10.1371/journal.pone.0155183 Shaw DE (2021) Desmond molecular dynamics system. MaestroDesmond interoperability tools. Research, Schrodinger Release Kumar S, Jayan J, Manoharan A, Benny F, Abdelgawad MA, Ghoneim MM, El-Sherbiny M, Sudevan T, Aneesh S, T.P. and, Mathew B (2023) Discerning of isatin-based monoamine oxidase (MAO) inhibitors for neurodegenerative disorders by exploiting 2D, 3D-QSAR modelling and molecular dynamics simulation. J Biomol Struct Dynamics, pp.1–13 Manoharan A, Oh JM, Benny F, Kumar S, Abdelgawad MA, Ghoneim MM, Shaker ME, El-Sherbiny M, Almohaimeed HM, Gahtori P, Kim H (2023) Assembling a Cinnamyl Pharmacophore in the C3-Position of Substituted Isatins via Microwave-Assisted Synthesis: Development of a New Class of Monoamine Oxidase-B Inhibitors for the Treatment of Parkinson’s Disease. Molecules, 28(16), p.6167 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7704948","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":526256711,"identity":"addbd492-297b-4160-b9d1-9acbdc7202e0","order_by":0,"name":"Roshni Pushpa Raghavan","email":"","orcid":"","institution":"Amrita School of Pharmacy, Amrita Vishwa Vidyapeetham","correspondingAuthor":false,"prefix":"","firstName":"Roshni","middleName":"Pushpa","lastName":"Raghavan","suffix":""},{"id":526256712,"identity":"108c8815-9b6d-40cf-b737-5566ba3b7d33","order_by":1,"name":"Muzammil Kabier","email":"","orcid":"","institution":"Amrita School of Pharmacy, Amrita Vishwa Vidyapeetham, AIMS Health Sciences Campus","correspondingAuthor":false,"prefix":"","firstName":"Muzammil","middleName":"","lastName":"Kabier","suffix":""},{"id":526256713,"identity":"8ba52d93-0ad1-4652-bd00-b7e5de583f41","order_by":2,"name":"Sunil Kumar","email":"","orcid":"","institution":"Amrita School of Pharmacy, Amrita Vishwa Vidyapeetham, AIMS Health Sciences Campus","correspondingAuthor":false,"prefix":"","firstName":"Sunil","middleName":"","lastName":"Kumar","suffix":""},{"id":526256716,"identity":"a16040da-e17d-44e3-a833-7bcea874f100","order_by":3,"name":"Kirti Theresa Alexander","email":"","orcid":"","institution":"Amrita School of Pharmacy, Amrita Vishwa Vidyapeetham","correspondingAuthor":false,"prefix":"","firstName":"Kirti","middleName":"Theresa","lastName":"Alexander","suffix":""},{"id":526256717,"identity":"ac68bd86-9ca8-4717-b547-d4d6fb1eb55b","order_by":4,"name":"Shine Sadasivan","email":"","orcid":"","institution":"Amrita Institute of Medical Sciences, AIMS Health Sciences Campus, Amrita Vishwa Vidyapeetham","correspondingAuthor":false,"prefix":"","firstName":"Shine","middleName":"","lastName":"Sadasivan","suffix":""},{"id":526256720,"identity":"e52260de-ee6c-40e3-b70e-383b84d1f00c","order_by":5,"name":"Mohamed A. Abdelgawad","email":"","orcid":"","institution":"Jouf University","correspondingAuthor":false,"prefix":"","firstName":"Mohamed","middleName":"A.","lastName":"Abdelgawad","suffix":""},{"id":526256724,"identity":"3fd67986-bdda-4b7b-9de2-30b9b9c22e47","order_by":6,"name":"Syed N. A. Bukhari","email":"","orcid":"","institution":"Jouf University","correspondingAuthor":false,"prefix":"","firstName":"Syed","middleName":"N. A.","lastName":"Bukhari","suffix":""},{"id":526256727,"identity":"2e52ff9e-ed06-479e-872d-9126986ea156","order_by":7,"name":"Bijo Mathew","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBUlEQVRIiWNgGAWjYBACA2aGBCBpAePbADFj4wEitEiA2IwNDAxpYBq/FggF13IYzMWrxZyd4eFjngIJeYPjvccffm07b7e2/TDQlhqbaFxaLJsZko15DCQMN5w5l9gs23Y7eduZRKCWY2m5DbgcdpghTXKGgQTjzBk5hs2SQC1mB4BaGBsOE9RiP3P+G5CWc8lm5x8S1iLxwUAisV+Cx7DxY9sBO7MbBGwB+cUAqCW5nyfHcDbDueQEsxtAWxLw+MWc/0zig4Q/NrZt7GcMPv4os7M3O5/+8MGHGhucWhgYeBLgTGYeBoZEsMoE7GqhgP0AnMn4g4HBHq/iUTAKRsEoGJEAANZ0YjsfA4xPAAAAAElFTkSuQmCC","orcid":"","institution":"Amrita School of Pharmacy, Amrita Vishwa Vidyapeetham, AIMS Health Sciences Campus","correspondingAuthor":true,"prefix":"","firstName":"Bijo","middleName":"","lastName":"Mathew","suffix":""}],"badges":[],"createdAt":"2025-09-24 14:53:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7704948/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7704948/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":93248431,"identity":"015be9aa-7fbf-4298-b9bd-e61053e75e57","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":930101,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript2492025.docx","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/4186d37f2c0661da91618a6b.docx"},{"id":93250440,"identity":"e8d76ebf-bb9a-411e-8d93-5035d5de1d4d","added_by":"auto","created_at":"2025-10-10 15:43:26","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9711,"visible":true,"origin":"","legend":"","description":"","filename":"d8abd7505c174b87b4408c0e11bdd12f.json","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/d5d0305a523080aa2b18b20a.json"},{"id":93250443,"identity":"45c067cc-9b27-45be-847b-3cdf0cb453fd","added_by":"auto","created_at":"2025-10-10 15:43:26","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":98673,"visible":true,"origin":"","legend":"","description":"","filename":"d8abd7505c174b87b4408c0e11bdd12f1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/eb31cce016f8e702017c2e2a.xml"},{"id":93250441,"identity":"584cf431-b93f-4065-bce8-8416d3fea594","added_by":"auto","created_at":"2025-10-10 15:43:26","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":209676,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/d3be3be87f16a42797ca5ee8.png"},{"id":93251685,"identity":"beeb93ac-a656-4863-82c5-a0eda923f464","added_by":"auto","created_at":"2025-10-10 15:51:26","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52070,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/251662de163bb0e284d1ac2c.png"},{"id":93251686,"identity":"5ab4880a-4d60-43f1-9142-1115cbddb637","added_by":"auto","created_at":"2025-10-10 15:51:26","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51397,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/41a16a775e12f1848d0d6273.png"},{"id":93248435,"identity":"43b1c56b-ac4f-4a9d-a0a9-6af29a0d69f8","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":169099,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/a8af902470bb14635f9ab897.jpeg"},{"id":93251690,"identity":"512772cf-fb33-4c8e-8d4c-05b124345496","added_by":"auto","created_at":"2025-10-10 15:51:26","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":201232,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/0ecff171e3c8601a5575eb9f.png"},{"id":93248442,"identity":"2fa6d786-4880-4d35-9c10-a830542931f9","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":134249,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/1ea1baf8a81e32ba6b53de13.png"},{"id":93250446,"identity":"bd8a5f21-9a1f-4728-a268-c520d6be3234","added_by":"auto","created_at":"2025-10-10 15:43:26","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":173550,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/463dee291a2fd3e59c4b468a.png"},{"id":93250450,"identity":"39d5dec5-c0dd-49ab-a04d-95ab3439a680","added_by":"auto","created_at":"2025-10-10 15:43:26","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":48054,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/2576728b05dfb16e69223770.png"},{"id":93248447,"identity":"cf152467-7693-4786-b03b-c33c7f19f816","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15506,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/6db47cf167e9b873f85e4251.png"},{"id":93248441,"identity":"98158a24-ef20-478e-931b-47f3422b0e11","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12835,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/45ec5ba09b9f736e55eed130.png"},{"id":93248448,"identity":"a512a3d8-439f-425c-b394-ee61363c3a05","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":31579,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/9f7309b9bc834ee73e0321c3.png"},{"id":93248445,"identity":"3b5d3c38-590b-4841-9cbc-59c42af6ec8c","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35184,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/fa218451952d00766c142d0d.png"},{"id":93252646,"identity":"09811e98-2ef3-40e8-a5dd-963699767fe3","added_by":"auto","created_at":"2025-10-10 15:59:26","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38687,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/d6a10fb351f8189a833ff368.png"},{"id":93251688,"identity":"08c113d3-8ae7-44a6-84b3-adfcb0124d0d","added_by":"auto","created_at":"2025-10-10 15:51:26","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":45884,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/0b802b1c3908fb65eb6c9ddd.png"},{"id":93250448,"identity":"ae1445a0-e136-4bfc-b81f-854d3a98d98d","added_by":"auto","created_at":"2025-10-10 15:43:26","extension":"xml","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":97116,"visible":true,"origin":"","legend":"","description":"","filename":"d8abd7505c174b87b4408c0e11bdd12f1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/b1d92e2b0921e23ffc95f7b4.xml"},{"id":93248450,"identity":"2244a347-d0a8-4b13-94cc-078345069093","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"html","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103779,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/1710c69f35ef649be4a3cb8f.html"},{"id":93248424,"identity":"ca163d1c-5a6d-4a55-ab94-f1b8d11b1b65","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":289140,"visible":true,"origin":"","legend":"\u003cp\u003et-SNE plot for each target (actives -red and inactives – blue)\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/dbe8b866b22301090a85e8cf.png"},{"id":93248426,"identity":"29228748-5bca-43e4-b2ee-54e99ae57c96","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":109220,"visible":true,"origin":"","legend":"\u003cp\u003eTanimoto similarity distribution bar plot for each target dataset, blue bar represents frequency of pair in specified bin of tanimoto similarity distribution. Red dashed line represents median of the distribution.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/5779d7ecf734c5d65c82aa8d.png"},{"id":93248428,"identity":"cdcfe31f-3354-42af-bb00-39daff64c29d","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":85216,"visible":true,"origin":"","legend":"\u003cp\u003eMCC Bar plot for each target, each bar represents different ML classifiers.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/148743ba41a6c675f02f3a30.png"},{"id":93248430,"identity":"b470a56c-46c2-405f-ba12-bf71ef412137","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":84031,"visible":true,"origin":"","legend":"\u003cp\u003eBox plot representing variance in minRMSD for 5 iterative redocking, each box represents each scoring function.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/bc0ac362b0dce5102b14be2e.png"},{"id":93248436,"identity":"0fbf44a6-7633-40ab-b967-5dc81a067bfb","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":162063,"visible":true,"origin":"","legend":"\u003cp\u003eSuperimposition of Re-docked and Native Pose. Cyan coloured ligand represents re-docked pose and Green coloured ligand represents native pose.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/6e77f61d954d435f57dc13b8.png"},{"id":93252645,"identity":"aac4bb7d-b8c3-491e-aa0b-c956d715e8ca","added_by":"auto","created_at":"2025-10-10 15:59:26","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":225086,"visible":true,"origin":"","legend":"\u003cp\u003eMolecular dynamics simulation analysis of 2AZ5, IKK, PDGF, and TGF complexes with active and inactive compounds. The plots represent the root-mean-square deviation (RMSD) profiles of the Cα atoms over the 100 ns simulation period, illustrating the structural stability and conformational fluctuations of each protein-ligand complex.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/651bec13222a29a0f41ca796.png"},{"id":93248437,"identity":"629d78f7-2a7b-436c-a285-2391a51c61bf","added_by":"auto","created_at":"2025-10-10 15:35:26","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":267220,"visible":true,"origin":"","legend":"\u003cp\u003eMolecular dynamics simulation analysis of 2AZ5, IKK, PDGF, and TGF protein complexes bound with active and inactive compounds. The root-mean-square fluctuation (RMSF) plots illustrate the flexibility of individual amino acid residues over the 100 ns simulation period, highlighting differences in residue-level mobility and structural dynamics between active and inactive ligand-bound states.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/ebdbef4a5c40fd6477667e7c.png"},{"id":94013879,"identity":"7e0f0c80-e36c-4e69-abf7-ac44245f9ead","added_by":"auto","created_at":"2025-10-21 10:42:24","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2035960,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7704948/v1/aa1573d8-9c29-4c6e-a966-f6248b2f7c10.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Smart Simulations for Scarred Livers: Target-Based Computational Models in Cirrhosis Care","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eCirrhosis is the end-stage of chronic liver disease, marked by extensive fibrosis, loss of liver function, and nodular regeneration. Fibrosis and inflammatory diseases represent major clinical challenges due to their chronic progression, complex pathophysiology, and limited treatment options.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e These conditions are often characterized by persistent immune activation, tissue remodeling, and extracellular matrix (ECM) accumulation, ultimately leading to organ dysfunction and failure. Traditional therapies typically offer symptomatic relief but fail to halt or reverse disease progression. In recent years, target-based approaches have gained momentum, aiming to intervene at specific molecular pathways implicated in disease pathogenesis\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Advances in genomics, proteomics, and bioinformatics have facilitated the identification of key targets such as cytokines, growth factors, and signaling molecules. \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThese targets contribute at multiple levels:TGF-β1 is the master regulator of fibrosis in cirrhosis. Its sustained activation results in collagen buildup and scar tissue formation\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. PDGF supports this by keeping stellate cells activated, contributing to fibrosis expansion.\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e IKKβ/NF-κB signaling maintains a chronic inflammatory environment, which causes continuous liver injury and HSC activation. TNF-α contributes to hepatocyte death and chronic inflammation, both of which are drivers of progressive liver damage and scarring. Therapies directed at transforming growth factor-beta (TGF-β), interleukins, and immune checkpoint regulators have shown promise in preclinical and early clinical studies\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. Moreover, small molecules, monoclonal antibodies, and RNA-based therapeutics are being developed to modulate these targets with increased specificity and reduced side effects. The integration of personalized medicine strategies further enhances the potential for targeted interventions by considering patient-specific molecular profiles\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Despite these advancements, challenges remain in translating molecular targets into effective, safe, and widely accessible treatments. This research explores the current landscape and future potential of target-based therapies for fibrosis and inflammatory diseases, highlighting both breakthroughs and barriers in the field. It gives clear explanation of the drug inhibitory efficacy across the four key protein targets \u0026mdash; TGF-β1, PDGF, IKKβ, and TNF-α how they are linked to inflammation, fibrosis, and specifically cirrhosis\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eClinical pharmacists play a crucial role in optimizing pharmacotherapy for Chronic diseases like Cirrhosis which navigates a landscape of emerging targeted therapies, biologics, and small molecules. However, the development of novel drugs is time-consuming and costly, necessitating approaches like artificial intelligence and machine learning which accelerate drug discovery and support the faster mode of clinical decision-making\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. In silico methods, particularly Quantitative Structure-Activity Relationship (QSAR) modeling and molecular docking, have emerged as powerful tools to predict compound activity against specific biological targets. These computational techniques allow rapid screening of large chemical libraries to identify potential drug candidates with favorable binding profiles. The research focuses on four key molecular targets implicated in fibrosis and inflammation: TGF-β1, PDGF, IKKB, and TNF-α. We integrate machine learning-based activity prediction and docking data to explore the pharmacotherapeutic potential of compounds targeting these pathways. Emphasis is placed on translating these findings into clinical pharmacy practice, highlighting how computational insights can guide drug selection, monitoring, and therapeutic optimization.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1. Computational Methodology\u003c/h2\u003e\u003cp\u003eBinary classification machine learning techniques were used to create the predictive models examined in this review, which evaluated drug inhibitory efficacy across four protein targets like TGF-β1, PDGF, IKKB, and TNF-α linked to inflammation and fibrosis.The datasets were curated from the ChEMBL database, comprising between 700 to 1300 compounds per target, following data cleaning procedures such as imputation, removal of duplicates, and chemical structure validation using RDKit \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e.A classification threshold of 500 nM (pIC50\u0026thinsp;=\u0026thinsp;6.3) was set to binarize compounds into active and inactive classes. Model development was proceeded with PoseidonQ software platform, utilizing Morgan fingerprints for molecular representation\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e different classifiers were benchmarked with 10-fold cross-validation, identifying Random Forest, HistGradientBoosting, and Support Vector Machine (SVM) as top performers for various targets. Models were internally validated using rigorous 10x10 K-fold cross-validation and externally validated with independent datasets to ensure robustness and generalizability. Applicability domains were defined based on Tanimoto similarity distributions to assess prediction confidence \u003csup\u003e\u003cspan additionalcitationids=\"CR12\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eComplementary molecular docking studies were performed using MzDOCK software to validate predicted compound-target interactions\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Docking protocols were calibrated through iterative redocking of co-crystal ligands to select optimal scoring functions, ensuring RMSD values below the accepted 2 \u0026Aring; threshold. Predicted active and inactive compound pairs were docked to their respective protein structures, confirming consistency between computational binding affinities and classification results. Together, these methods provide a strong computational foundation for predicting compound efficacy and informing pharmacotherapeutic decisions. Add the details of the MD simulation with procedure and references.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2. ML-Model Dataset, Development and Validatory Protocol\u003c/h2\u003e\u003cp\u003eA total of Binary Classification based QSAR models were developed, capturing inhibitory relationship in four targets consisting of a pair each from anti-fibrotic (TGF- β1 and PDGF) and anti-inflammatory (IKKB and TNF-α) pathways. Initial IC\u003csub\u003e50\u003c/sub\u003e dataset was extracted from ChEMBL database for each target as specified below:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eTGF- β1 - CHEMBL4439 (1018 Compounds)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003ePDGF - CHEMBL1913 (1285 Compounds)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIKKB - Merged CHEMBL3529 and CHEMBL1991 (1340 Compounds)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eTNF-α \u0026ndash; CHEMBL1825 (706 Compounds)\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eData imputation, duplicate entry removal and RDKit check were performed to clean the datasets. Random 20% dataset was extracted from the Initial dataset for external validation. The Binary classification threshold was set to 500 nM (pIC50\u0026ndash;6.30) for binarization of y label (0: inactive and 1: active). A t-SNE plot was generated for visualization of actives/inactives in chemical space to observe isolated cluster formation.\u003c/p\u003e\u003cp\u003eApplicability domain for each target were defined on the basis of pairwise tanimoto similarity distribution with Morgan Fingerprint upon which median is determined to assign the threshold for mean tanimoto similarity between 5 nearest compounds in space corresponding to the predicted compound. Comparative analysis of 11 Classification model to identify the best model for each target dataset on the basis of Mathews Correlation Coefficient (MCC) was performed with K-Fold CV set to 10 iteration and the bar plot was produced for comparative analysis. The best model was used to build the model with default parameters for each target with 10x10 K-Fold Cross validation using Morgan fingerprint generating internal validatory metrics (Accuracy, Sensitivity, Specificity and MCC). Furthermore, to determine the true predictive nature, an external validation was conducted with extracted external dataset.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e\u003cb\u003e2.3. Deployment of Multi-Model System into Streamlit Cloud\u003c/b\u003e\u003c/h2\u003e\u003cp\u003eThe validated multi model system was deployed into streamlit cloud in the form of web application where user can input single SMILES, multiple SMILES through Excel sheet or sketch molecule with the help of chemical sketcher. The web application predicts whether the compound is active or inactive with respect to applicability domain. The threshold of AD is by default the median of tanimoto similarity distribution but is made to be flexible to impose strict applicability domain. The web application can be utilized in the link: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://mutli-model-cirrhosis.streamlit.app/\u003c/span\u003e\u003cspan address=\"https://mutli-model-cirrhosis.streamlit.app/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4. Molecular Docking Classification of Predicted Acitves/Inactives\u003c/h2\u003e\u003cp\u003ePredicted single pair of actives and inactives from each target model were docked to respective protein targets to identify whether the model prediction could align with binding scores. For each target, protein structure with co-crystal ligand was extracted from RCSB PDB database which is given below\u003csup\u003e\u003cspan additionalcitationids=\"CR16 CR17\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eTGF- β1\u0026ndash;1PY5 (co-crystal inhibitor resname: PY1)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003ePDGF \u0026ndash; 6JOL (co-crystal inhibitor resname: STI)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIKKB \u0026ndash; 4KIK (co-crystal inhibitor resname: KSA)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eTNF-α \u0026ndash; 2AZ5 (co-crystal inhibitor resname: 307)\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eMolecular docking study was conducted with software MzDOCK. The single pair of predicted actives and inactives from each target models were prepared using MMFF94 forcefield in protonation state of pH 7.4\u003csup\u003e19\u003c/sup\u003e. The respective protein of each target was prepared by adding polar hydrogens, assigning kollman charges and removing heteroatoms\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. The binding site was configured at the site of co-crystal ligand of each protein targets with a buffer-space of 4 \u0026Aring;. The choice of scoring function was determined by 5 iterations of redocking to identify the scoring function which produces lower RMSD between the redocked and native pose\u003csup\u003e\u003cspan additionalcitationids=\"CR22\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. A box plot representing minimum RMSD deviation in each scoring function for each target was generated. the num_modes and exhaustiveness were set to 9 and 8 respectively and the active\\inactive pair were docked with best scoring function. The redocking RMSD and superimposition of redocked and native pose for each target was performed for validatory purpose. The binding affinity of random active inactive pair was tabulated.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Molecular Dynamic simulation\u003c/h2\u003e\u003cp\u003eMolecular dynamics (MD) simulations were performed to examine the impact of the solvent environment on the structural stability of the protein\u0026ndash;ligand complex using the Desmond v7.2 package (Schr\u0026ouml;dinger Release 2022-4)\u003csup\u003e24\u003c/sup\u003e. The simulations were done on a Dell Precision 7820 Tower workstation running Ubuntu 22.04.1 LTS (64-bit) and outfitted with an Intel Xeon Silver 4210R processor and an NVIDIA RTX A4000 GPU. For all MD calculations, the OPLS4 force field was used. SPC water molecules were used to solvate the protein\u0026ndash;ligand complex, which was placed in the middle of an orthorhombic simulation box with a buffer of 10 \u0026Aring; between the protein and the box's edges. Counterions (Na⁺ and Cl⁻) were added at random to neutralize the simulation system, and the complex's shape was used to modify the periodic boundary box's volume. The dynamic behaviour of the complex was evaluated by running the 100 ns MD simulation under NPT conditions. The conformational stability and domain-specific interactions of the complex were assessed employing all Cα atoms in key structural studies, such as protein\u0026ndash;ligand interaction profiling, root mean square deviation (RMSD), and root mean square fluctuation (RMSF)\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Results and Discussion","content":"\u003cp\u003e\u003cb\u003e3.1. t-SNE Plot for Active/Inactive Cluster Visualization\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAn important observation is that actives form isolated clusters and inactives are sparse, which affects the specificity of the model but there is an exception in the case of TNF- α were inactives are forming clusters\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.2. Pairwise Tanimoto Similarity Distribution Plot\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFor all 4 targets, the median is low indicating highly sparse and low similarity heterogenous dataset (Fig.\u0026nbsp;2).\u003c/p\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Comparative Analysis of 11 Classification Models\u003c/h2\u003e\u003cp\u003eRandom Forrest classifier performed best for TGF-beta and IKKB with MCC of 0.65\u0026thinsp;\u0026plusmn;\u0026thinsp;0.087 and 0.69\u0026thinsp;\u0026plusmn;\u0026thinsp;0.051 respectively, HistGradientBoosting classifier for PDGF target with MCC of 0.64\u0026thinsp;\u0026plusmn;\u0026thinsp;0.060 and SVM classifier for TNF-α with MCC of 0.63\u0026thinsp;\u0026plusmn;\u0026thinsp;0.11 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Internal Validation of 4 Models\u003c/h2\u003e\u003cp\u003eInternal validation demonstrates that all the 4 models across the 10x10 Folds achieved over 0.6 MCC which indicates good performance and an accuracy of over 80%. TGF- β1 model has low specificity due to sparse inactive (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eInternal Validation Metrics for Each Target After 10x10 K-Fold CV\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTarget\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMCC\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTGF-β1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e0.880\u0026thinsp;\u0026plusmn;\u0026thinsp;0.032\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e0.965\u0026thinsp;\u0026plusmn;\u0026thinsp;0.024\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e0.585\u0026thinsp;\u0026plusmn;\u0026thinsp;0.123\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e\u003cp\u003e0.629\u0026thinsp;\u0026plusmn;\u0026thinsp;0.107\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePDGF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e0.834\u0026thinsp;\u0026plusmn;\u0026thinsp;0.038\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e0.887\u0026thinsp;\u0026plusmn;\u0026thinsp;0.042\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e0.748\u0026thinsp;\u0026plusmn;\u0026thinsp;0.072\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e\u003cp\u003e0.645\u0026thinsp;\u0026plusmn;\u0026thinsp;0.078\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIKKB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e0.843\u0026thinsp;\u0026plusmn;\u0026thinsp;0.033\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e0.830\u0026thinsp;\u0026plusmn;\u0026thinsp;0.052\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e0.858\u0026thinsp;\u0026plusmn;\u0026thinsp;0.052\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e\u003cp\u003e0.688\u0026thinsp;\u0026plusmn;\u0026thinsp;0.065\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTNF-α\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e0.834\u0026thinsp;\u0026plusmn;\u0026thinsp;0.046\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e0.755\u0026thinsp;\u0026plusmn;\u0026thinsp;0.080\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e0.880\u0026thinsp;\u0026plusmn;\u0026thinsp;0.053\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e\u003cp\u003e0.641\u0026thinsp;\u0026plusmn;\u0026thinsp;0.101\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.5. External Validation of 4 Models\u003c/h2\u003e\u003cp\u003eThe difference between the internal and external validation is minimal indicating robustness of the models and its ability to predict unseen data (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eExternal Validation Metrics with Random 20% External Dataset\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTarget\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy_ext\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSensitivity_ext\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSpecificity_ext\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMCC_ext\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTGF-β1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.892\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.955\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.688\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.687\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePDGF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.790\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.911\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.631\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.574\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIKKB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.840\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.832\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.849\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.678\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTNF-α\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.796\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.723\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.832\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.547\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Iterative Redocking with 6 Scoring function for Determination of Ideal Scorer\u003c/h2\u003e\u003cp\u003eFrom the plot, Vina performed better in 6JOL and 2AZ5 and Vinardo performed better in 1PY5 and 4KIK (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e4\u003c/span\u003e). This scoring function were selected for docking due to the precise local optimization of poses close to global minimum.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.5. Redocking of Co-Crystal Ligand and Superimposition\u003c/h2\u003e\u003cp\u003eTo validate the docking protocol, re-docking of the co-crystallized ligands into their respective protein targets was performed, and the RMSD values between the re-docked and native poses were calculated. The RMSD values obtained were 0.353 \u0026Aring; for 1PY5, 0.388 \u0026Aring; for 6JOL, 0.728 \u0026Aring; for 4KIK, and 0.538 \u0026Aring; for 2AZ5 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e5\u003c/span\u003e). All values were below the acceptable threshold of 2.0 \u0026Aring;, confirming the reliability of the docking procedure. The superimposition of the re-docked and native ligand poses is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e5\u003c/span\u003e, where the cyan-colored ligand represents the re-docked pose and the green-colored ligand represents the native crystallographic pose. The close overlap between the two poses demonstrates the robustness and accuracy of the docking methodology used in this study.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e3.6. Binding Affinity of Predicted Actives/Inactive\u003c/h2\u003e\u003cp\u003eMolecular docking was performed to compare the binding affinities of predicted active and inactive compounds against each target protein. The results are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. For 1PY5, the predicted active compound showed a binding affinity of \u0026minus;\u0026thinsp;10.3 kcal/mol, which was stronger than the inactive compound (\u0026ndash;8.4 kcal/mol) and close to the co-crystal ligand (\u0026ndash;8.8 kcal/mol). For 6JOL, the active compound exhibited the highest binding affinity (\u0026ndash;13.6 kcal/mol), consistent with the co-crystal ligand (\u0026ndash;13.6 kcal/mol), whereas the inactive compound showed a significantly weaker and unstable binding score (+\u0026thinsp;27.8 kcal/mol), indicating poor binding. In the case of 4KIK, the predicted active ligand had a binding affinity of \u0026minus;\u0026thinsp;10.9 kcal/mol, outperforming the inactive (\u0026ndash;8.6 kcal/mol), but slightly lower than the co-crystal ligand (\u0026ndash;13.0 kcal/mol).For 2AZ5, the active compound bound with a score of \u0026minus;\u0026thinsp;9.4 kcal/mol, which was stronger than the inactive (\u0026ndash;7.7 kcal/mol) and identical to the co-crystal ligand (\u0026ndash;9.4 kcal/mol). Overall, the docking results showed that the predicted active compounds consistently displayed stronger binding affinities compared to the inactives across all targets, further validating their potential as effective inhibitors.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eBinding Affinity of predicted actives\\inactive of each target\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eTargets\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eBinding Affinity\u003c/p\u003e\u003cp\u003e(kcal/mol)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003e1PY5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eActives\u003c/b\u003e\u003c/p\u003e\u003cp\u003eCC1\u0026thinsp;=\u0026thinsp;NC(=\u0026thinsp;CC\u0026thinsp;=\u0026thinsp;C1)C2\u0026thinsp;=\u0026thinsp;NN3CCCC3\u0026thinsp;=\u0026thinsp;C2C4\u0026thinsp;=\u0026thinsp;C5C\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;CC5\u0026thinsp;=\u0026thinsp;NC\u0026thinsp;=\u0026thinsp;C4)C(=\u0026thinsp;O)N\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-10.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eInactive\u003c/b\u003e\u003c/p\u003e\u003cp\u003eCOCC1c2c(-c3ccncc3)c(-c3cccc(C)n3)nn2CC1(C)C\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-8.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eco-crystal (redock)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-8.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003e6JOL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eActives\u003c/b\u003e\u003c/p\u003e\u003cp\u003eCC1\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;C1)NC(=\u0026thinsp;O)C2\u0026thinsp;=\u0026thinsp;CC\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;C2)CN3CCN(CC3)C)NC4\u0026thinsp;=\u0026thinsp;NC\u0026thinsp;=\u0026thinsp;CC(=\u0026thinsp;N4)C5\u0026thinsp;=\u0026thinsp;CN\u0026thinsp;=\u0026thinsp;CC\u0026thinsp;=\u0026thinsp;C5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-13.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eInactive\u003c/b\u003e\u003c/p\u003e\u003cp\u003eO\u0026thinsp;=\u0026thinsp;C1N/C\u0026thinsp;=\u0026thinsp;C/c2ccc(c(Br)c2)Oc2cc(cc(Br)c2O)CCNC(=\u0026thinsp;O)/C(=\u0026thinsp;N/O)Cc2ccc(c(Br)c2)Oc2cc(cc(Br)c2O)C/C1\u0026thinsp;=\u0026thinsp;N\\O\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e27.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eco-crystal (redock)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-13.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003e4KIK\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eActives\u003c/b\u003e\u003c/p\u003e\u003cp\u003eC1COCCC1OC2\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;C2)C3\u0026thinsp;=\u0026thinsp;NC(=\u0026thinsp;NC\u0026thinsp;=\u0026thinsp;N3)NC4\u0026thinsp;=\u0026thinsp;CC\u0026thinsp;=\u0026thinsp;C(C\u0026thinsp;=\u0026thinsp;C4)N5CCN(CC5)C6COC6)C#N\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-10.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eInactive\u003c/b\u003e\u003c/p\u003e\u003cp\u003eCc1noc(-c2ccc3nc(Nc4cc(CN5CCN(CC(F)F)CC5)ccn4)[nH]c3c2)n1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-8.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eco-crystal (redock)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-13\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003e2AZ5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eActives\u003c/b\u003e\u003c/p\u003e\u003cp\u003eCOc1ccc2c(c1F)C(=\u0026thinsp;O)N(C[C@@]1(C#Cc3ccc(-c4nc(-c5cnn(C)c5)ccc4OC(=\u0026thinsp;O)C(C)(C)C)c(F)c3)NC(=\u0026thinsp;O)NC1\u0026thinsp;=\u0026thinsp;O)C2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-9.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eInactive\u003c/b\u003e\u003c/p\u003e\u003cp\u003eCCCn1c(=\u0026thinsp;O)n(CC)c(=\u0026thinsp;O)c2cnc3c(OC)cccc3c21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-7.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eco-crystal (redock)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-9.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.7. Molecular Dynamics\u003c/b\u003eThe RMSD values for each target protein in complex with the active and inactive compounds were as follows:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e2AZ5: The complex with the active compound exhibited an average RMSD of 3.69 \u0026Aring; (range: 1.36\u0026ndash;4.59 \u0026Aring;), while the complex with the inactive compound showed a slightly higher average RMSD of 4.09 \u0026Aring; (range: 1.39\u0026ndash;5.40 \u0026Aring;).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIKK: The active compound complex had an average RMSD of 4.82 \u0026Aring; (range: 1.74\u0026ndash;6.87 \u0026Aring;), whereas the inactive compound complex had an average RMSD of 4.25 \u0026Aring; (range: 1.78\u0026ndash;6.23 \u0026Aring;).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003ePDGF: The complex with the active compound maintained an average RMSD of 2.00 \u0026Aring; (range: 1.02\u0026ndash;2.54 \u0026Aring;), while the inactive compound complex had a slightly higher RMSD of 2.17 \u0026Aring; (range: 0.94\u0026ndash;2.79 \u0026Aring;).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eTGF: The active compound complex showed an average RMSD of 1.90 \u0026Aring; (range: 1.16\u0026ndash;2.52 \u0026Aring;), compared to 2.07 \u0026Aring; (range: 1.15\u0026ndash;2.62 \u0026Aring;) for the inactive compound complex.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThese results suggest that, across all targets, complexes with active compounds generally exhibited slightly lower or more stable RMSD profiles, supporting stronger and more consistent binding interactions compared to their inactive counterparts. In addition to RMSD analysis, root mean square fluctuation (RMSF) was employed to assess the flexibility of specific residues in the protein-ligand complexes. RMSF provides critical insights into residue-level mobility, where lower values suggest stable, well-ordered regions often corresponding to secondary structures such as α-helices and β-sheets while higher values typically reflect flexible regions such as loops and terminal residues. Lower RMSF values at binding site residues indicate stronger ligand-protein interactions and conformational stability. For the 2AZ5 complex, the active compound induced lower fluctuations in key binding site residues such as Val13 (0.73 \u0026Aring;), His15 (0.65 \u0026Aring;), Tyr59 (0.77 \u0026Aring;), and Val123 (0.71 \u0026Aring;). In contrast, the inactive compound complex showed elevated fluctuations for comparable residues, such as Tyr59 (1.093 \u0026Aring;), Gly121 (1.03 \u0026Aring;), And Gly122 (1.57 \u0026Aring;), suggesting reduced binding stability. In the IKK complex, residues interacting with the active compound, including Glu19 (3.48 \u0026Aring;), Thr23 (3.52 \u0026Aring;), and Gly27 (3.65 \u0026Aring;), showed higher flexibility but were consistently involved in ligand interactions. Comparatively, the inactive compound complex exhibited lower RMSF values for these residues (Glu19 at 2.11 \u0026Aring; and Thr23 at 1.93 \u0026Aring;), suggesting a more rigid but potentially less interactive binding conformation. The PDGF complex with the active compound displayed moderate fluctuations in residues such as Ala640 (1.71 \u0026Aring;), Ser643 (1.80 \u0026Aring;), and Glu644 (1.89 \u0026Aring;), indicating dynamic but sustained interactions. In contrast, the inactive compound yielded much lower RMSF values for analogous residues, such as Glu644 (0.75 \u0026Aring;) and Leu_651 (1.02 \u0026Aring;), suggesting reduced mobility and potentially weaker engagement. For TGF, the active compound complex showed higher fluctuations in residues like Gly214 (2.37 \u0026Aring;), His285 (1.5 \u0026Aring;), And Glu284 (1.28 \u0026Aring;), reflecting active engagement with the binding site. The inactive compound displayed significantly reduced fluctuations in these regions, such as His285 (1.11 \u0026Aring;) and Gly286 (0.81 \u0026Aring;), indicating less dynamic interaction with the protein core. These RMSF trends highlight a consistent pattern: complexes with active compounds tend to show moderate residue flexibility around the binding site, reflecting both stability and adaptability, whereas inactive compound complexes display either excessive rigidity or limited movement, possibly leading to suboptimal binding and interaction profiles. These observations corroborate the superior binding characteristics of the active ligands and support their candidacy for further optimization. The RMSD and RMSF plots are given in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e6\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e7\u003c/span\u003e respectively. A Table summarizing RMSF analysis of active and inactive in each target is given in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary of RMSF Analysis for Active and Inactive Compounds Across Target Proteins.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTarget Protein\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCompound Type\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eKey Residues (Selected)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eRMSF Range (\u0026Aring;)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eNotable Observations\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003e2AZ5\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eActive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eVal13, His15, Tyr59, Val123, Ile155\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.55\u0026ndash;1.52\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eLower fluctuations at active site; stable secondary structures retained\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eInactive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTyr59, Gly121, Gly122, Lys11, Leu_157\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.76\u0026ndash;2.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eHigher fluctuations in active site residues; potential loss of stable interactions\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eIKK\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eActive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGlu19, Thr23, Gly27, Tyr98, Gly168\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.29\u0026ndash;5.67\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eModerate-to-high fluctuations; sustained interactions at flexible active site\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eInactive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGlu19, Thr23, Tyr98, Asp103, Glu149\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.98\u0026ndash;2.56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eLower flexibility but possibly weaker ligand accommodation\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003ePDGF\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eActive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAla640, Ser643, Glu644, His650, Val_815\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.61\u0026ndash;1.89\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eBalanced flexibility; engagement of key residues via adaptable side chains\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eInactive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSer643, Glu644, Leu651, Cys814, Val815\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.43\u0026ndash;1.02\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMore rigid binding site; less conformational adaptability\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eTGF\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eActive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGly214, His285, Glu284, Asp290\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.5\u0026ndash;2.37\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eDynamic interaction at core residues; high engagement flexibility\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eInactive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHis285, Arg294, Gly286, Asp351\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.5\u0026ndash;1.31\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eRelatively stable but lower RMSF flexibility; potentially suboptimal binding dynamics\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eAn interesting observation from the internal validation metrics of the models given in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e is MCC of all the target models is above 0.6 indicating good predictive nature, considering a rigorous validation of 10x10 K-fold CV. High sensitivity is seen in TGF-β1 due to isolated active clusters, but poor specificity due to sparse inactives in chemical space. The specificity in TNF-α is particularly high due to clustered inactives. Although, all target models had MCC of above 0.6, TGF-β1 and TNF-α shows higher standard deviation due to its sparse active\\inactive data distribution nature. External validation shows that all the target models have good predictive capability considering the dataset was highly heterogenous with natural class distribution imbalances. MCC_ext and ACC_ext of above 0.54 and 0.79 respectively as given in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e within AD definition indicates the model has good predictive power in unseen data. The deviation in MCC between internal and external metric is minimal which establish the robustness of the 4 QSAR models. For further validation, molecular docking studies were conducted with proper redocking protocol beforehand to calibrate the conditions to produce high resolution poses. RMSD of redocking protocol was observed to be below the common threshold of \u0026lt;\u0026thinsp;2 \u0026Aring; which indicates the scoring function is well adapted to find global minimum. The docking was conducted in the same seed of redocking procedure and the results are given in results section Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. From the observation of Table\u0026nbsp;4, the binding score of the random pairs of predicted actives and inactives in each target were successfully discriminated by the chosen scoring function, aligning with the model prediction results. This provides evidence of model\u0026rsquo;s predictive expansion within biological context. Furthermore, the MD simulations provided a much more effective evidence that the prediction of the model could be even be translated in to the compounds effect in the kinetics of the protein.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe application of QSAR, molecular docking and dynamics targeting TGF-β1, PDGF, IKKB, and TNF-α underscores the potential of computational methods to revolutionize pharmacotherapy in fibrosis and inflammatory diseases. These approaches identify promising therapeutic candidates and provide a robust framework for clinical pharmacists to enhance drug selection, monitoring, and patient care. They bridge the gap between drug discovery and clinical application, ensuring that novel compounds targeting fibrogenic pathways are evaluated for safety, efficacy, and drug interactions. Furthermore, their expertise in pharmacokinetics and pharmacodynamics enables the optimization of personalized therapies for cirrhosis patients, ultimately improving treatment outcomes\u003c/p\u003e\u003cp\u003eThe provided validated multi-model system with defined applicability domain within a single web-based platform can be of great utility to researchers. The models are reliable and rigorously validated and hence can be used in research application by the scientific community from the given link: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://mutli-model-cirrhosis.streamlit.app/\u003c/span\u003e\u003cspan address=\"https://mutli-model-cirrhosis.streamlit.app/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eThe Deanship of Graduate Studies and Scientific Research at Jouf University under grant No. (DGSSR-2025-01-01639)\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eR.P.R. contributed to manuscript writing. S.S. manuscript writing. K.T.A. performed clinical writing and manuscript writing. M. K. and S.K. carried out software development, molecular simulations, computational analysis, and manuscript writing. S.K. and B.M. contributed to conceptualization and supervision. B.M. also provided medicinal chemistry insights, validation, editing, and drafting. M.A.A. and S.N.A.B. were involved in overall supervision and project administration. All authors reviewed and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThis work was funded by the Deanship of Graduate Studies and Scientific Research at Jouf University under grant No. (DGSSR-2025-01-01639)\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSchuppan D, Afdhal NH (2008) Liver cirrhosis. Lancet 371(9615):838\u0026ndash;851. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0140-6736(08)60383-9\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(08)60383-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHopkins AL (2008) Network pharmacology: the next paradigm in drug discovery. Nat Chem Biol 4(11):682\u0026ndash;690. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nchembio.118\u003c/span\u003e\u003cspan address=\"10.1038/nchembio.118\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRifai N, Gillette MA, Carr SA (2006) Protein biomarker discovery and validation: the long and uncertain path to clinical utility. Nat Biotechnol 24(8):971\u0026ndash;983. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nbt1235\u003c/span\u003e\u003cspan address=\"10.1038/nbt1235\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMeng XM, Nikolic-Paterson DJ, Lan HY (2016) TGF-β: the master regulator of fibrosis. Nat Rev Nephrol 12(6):325\u0026ndash;338. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrneph.2016.48\u003c/span\u003e\u003cspan address=\"10.1038/nrneph.2016.48\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBorkham-Kamphorst E, Kovalenko E, van Roeyen C-R, Gassler N, Bomble M, Ostendorf T, Floege J, Gressner AM, Weiskirchen R (2008) Platelet-derived growth factor isoform expression in carbon tetrachloride-induced chronic liver injury. Lab Invest 88(10):1090\u0026ndash;1100. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/labinvest.2008.71\u003c/span\u003e\u003cspan address=\"10.1038/labinvest.2008.71\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWirth T, Adler G et al (2012) Hepatic activation of IKK/NF-κB signaling induces liver fibrosis via macrophage-mediated chronic inflammation. Hepatology 55(3):1037\u0026ndash;1045. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/hep.24721\u003c/span\u003e\u003cspan address=\"10.1002/hep.24721\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBradham CA, Pl\u0026uuml;mpe J, Manns MP, Brenner DA, Trautwein C (1998) I. TNF-induced liver injury. Am J Physiol Gastrointest Liver Physiol 275(3):G387\u0026ndash;G392. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1152/ajpgi.1998.275.3.G387\u003c/span\u003e\u003cspan address=\"10.1152/ajpgi.1998.275.3.G387\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMendez D, Gaulton A, Bento AP, Chambers J, De Veij M, F\u0026eacute;lix E, Magari\u0026ntilde;os MP, Mosquera JF, Mutowo P, Nowotka M, Gordillo-Mara\u0026ntilde;\u0026oacute;n M, Hunter F, Junco L, Mugumbate G, Rodriguez-Lopez M, Atkinson F, Bosc N, Radoux CJ, Segura-Cabrera A, Hersey A, Leach AR (2019) ChEMBL: towards direct deposition of bioassay data. Nucleic Acids Res 47(D1):D930\u0026ndash;D940. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gky1075\u003c/span\u003e\u003cspan address=\"10.1093/nar/gky1075\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRDKit Open-source cheminformatics. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.rdkit.org\u003c/span\u003e\u003cspan address=\"https://www.rdkit.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKabier M, Gambacorta N, Ciriaco F, Mastrolorito F, Kumar S, Mathew B, Nicolotti O (2025) PoseidonQ: A Free Machine Learning Platform for the Development, Analysis, and Validation of Efficient and Portable QSAR Models for Drug Discovery. J Chem Inf Model 65(8):3944\u0026ndash;3954. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/acs.jcim.4c02372\u003c/span\u003e\u003cspan address=\"10.1021/acs.jcim.4c02372\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBajusz D, R\u0026aacute;cz A, H\u0026eacute;berger K (2015) Why Is Tanimoto Index an Appropriate Choice for Fingerprint-Based Similarity Calculations? J Cheminform 7(1):20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13321-015-0069-3\u003c/span\u003e\u003cspan address=\"10.1186/s13321-015-0069-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCiriaco F, Gambacorta N, Alberga D, Nicolotti O (2021) Quantitative Polypharmacology Profiling Based on a Multifingerprint Similarity Predictive Approach. J Chem Inf Model 61(10):4868\u0026ndash;4876. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1021/acs.jcim.1c00498\u003c/span\u003e\u003cspan address=\"10.1021/acs.jcim.1c00498\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCiriaco F, Gambacorta N, Trisciuzzi D, Nicolotti OPLATO (2022) A Predictive Drug Discovery Web Platform for Efficient Target Fishing and Bioactivity Profiling of Small Molecules. Int J Mol Sci 23(9):5245. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ijms23095245\u003c/span\u003e\u003cspan address=\"10.3390/ijms23095245\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKabier M, Gambacorta N, Trisciuzzi D, Kumar S, Nicolotti O, Mathew B (2024) MzDOCK: A free ready-to-use GUI-based pipeline for molecular docking simulations. J Comput Chem 45(23):1980\u0026ndash;1986. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jcc.27390\u003c/span\u003e\u003cspan address=\"10.1002/jcc.27390\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBerman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN, Bourne PE (2000) The Protein Data Bank. Nucleic Acids Res 28(1):235\u0026ndash;242. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/28.1.235\u003c/span\u003e\u003cspan address=\"10.1093/nar/28.1.235\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHe MM, Smith AS, Oslob JD, Flanagan WM, Braisted AC, Whitty A, Cancilla MT, Wang J, Lugovskoy AA, Yoburn JC, Fung AD, Farrington G, Eldredge JK, Day ES, Cruz LA, Cachero TG, Miller SK, Friedman JE, Choong IC, Cunningham BC (2005) Small-molecule inhibition of TNF-alpha. Science 310(5750):1022\u0026ndash;1025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.1116304\u003c/span\u003e\u003cspan address=\"10.1126/science.1116304\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu S, Misquitta YR, Olland A, Johnson MA, Kelleher KS, Kriz R, Lin LL, Stahl M, Mosyak L (2013) Crystal structure of a human IκB kinase β asymmetric dimer. J Biol Chem 288(31):22758\u0026ndash;22767. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1074/jbc.M113.482596\u003c/span\u003e\u003cspan address=\"10.1074/jbc.M113.482596\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSawyer JS, Beight DW, Britt KS, Anderson BD, Campbell RM, Goodson T Jr, Herron DK, Li HY, McMillen WT, Mort N, Parsons S, Smith EC, Wagner JR, Yan L, Zhang F, Yingling JM (2004) Synthesis and activity of new aryl- and heteroaryl-substituted 5,6-dihydro-4H-pyrrolo[1,2-b]pyrazole inhibitors of the transforming growth factor-beta type I receptor kinase domain. Bioorg Med Chem Lett 14(13):3581\u0026ndash;3584. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.bmcl.2004.04.007\u003c/span\u003e\u003cspan address=\"10.1016/j.bmcl.2004.04.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHalgren TA, MMFF VII (1999) Characterization of MMFF94, MMFF94s, and other widely available force fields for conformational energies and for intermolecular-interaction energies and geometries. J Comput Chem 20(7):730\u0026ndash;748. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/(SICI)1096-987X(199905)20:7%3C730::AID-JCC8%3E3.0.CO;2-T\u003c/span\u003e\u003cspan address=\"10.1002/(SICI)1096-987X(199905)20:7%3C730::AID-JCC8%3E3.0.CO;2-T\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Wolf RM, Caldwell JW, Kollman PA, Case DA (2004) Development and testing of a general amber force field. J Comput Chem. ;25(9):1157-74. doi: 10.1002/jcc.20035. Erratum in: J Comput Chem. 2005;26(1):114\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKoes DR, Baumgartner MP, Camacho CJ (2013) Lessons learned in empirical scoring with smina from the CSAR 2011 benchmarking exercise. J Chem Inf Model 53(8):1893\u0026ndash;1904. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/ci300604z\u003c/span\u003e\u003cspan address=\"10.1021/ci300604z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEberhardt J, Santos-Martins D, Tillack AF, Forli S, AutoDock (2021) Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model 61(8):3891\u0026ndash;3898. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/acs.jcim.1c00203\u003c/span\u003e\u003cspan address=\"10.1021/acs.jcim.1c00203\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eQuiroga R, Villarreal MA (2016) Vinardo: A Scoring Function Based on Autodock Vina Improves Scoring, Docking, and Virtual Screening. PLoS ONE 11(5):e0155183. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0155183\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0155183\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShaw DE (2021) Desmond molecular dynamics system. MaestroDesmond interoperability tools. Research, Schrodinger Release\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKumar S, Jayan J, Manoharan A, Benny F, Abdelgawad MA, Ghoneim MM, El-Sherbiny M, Sudevan T, Aneesh S, T.P. and, Mathew B (2023) Discerning of isatin-based monoamine oxidase (MAO) inhibitors for neurodegenerative disorders by exploiting 2D, 3D-QSAR modelling and molecular dynamics simulation. J Biomol Struct Dynamics, pp.1\u0026ndash;13\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eManoharan A, Oh JM, Benny F, Kumar S, Abdelgawad MA, Ghoneim MM, Shaker ME, El-Sherbiny M, Almohaimeed HM, Gahtori P, Kim H (2023) Assembling a Cinnamyl Pharmacophore in the C3-Position of Substituted Isatins via Microwave-Assisted Synthesis: Development of a New Class of Monoamine Oxidase-B Inhibitors for the Treatment of Parkinson\u0026rsquo;s Disease. Molecules, 28(16), p.6167\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"QSAR, Drug Discovery, Cirrhosis, Molecular Docking, Molecular Dynamics","lastPublishedDoi":"10.21203/rs.3.rs-7704948/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7704948/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRecent advances in computational drug discovery, including Quantitative Structure-Activity Relationship (QSAR) modelling, molecular docking and molecular dynamics, have paved the way for identifying novel therapeutic candidates targeting key pathways involved in fibrosis and chronic inflammation. The present research retrieved the data from ChEMBL database to develop the binary classification-based machine learning models for four critical targets: Transforming Growth Factor Beta 1 (TGF-β1), Platelet-Derived Growth Factor (PDGF), Inhibitor of Nuclear Factor Kappa-B Kinase Subunit Beta (IKKB), and Tumor Necrosis Factor Alpha (TNF-α). These targets are implicated in diseases such as liver fibrosis, non-alcoholic steatohepatitis (NASH), and Cirrhosis. We discuss the pharmacotherapeutic relevance of predicted active compounds, their mechanisms of action, and implications for clinical pharmacy practice, including drug monitoring and therapy optimization.The study demonstrates how clinical pharmacists may navigate experimental discoveries with practical therapeutic approaches by using computational predictions to inform drug selection, repurposing, and customized patient care.The four machine learning models, from each selected through comparative analysis between 11 classifiers were rigorously validated with 10x10 K-fold cross validation and achieves MCC_ext\u0026thinsp;\u0026ge;\u0026thinsp;0.54 and Accuracy_ext\u0026thinsp;\u0026ge;\u0026thinsp;0.74 indicating robust predictive ability. Target based molecular simulations were conducted to chosen pairs of predicted actives and inactives in each target to further validate the discriminatory potential of the model from a simulation point of view and was found to be consistent with the binding scores and dynamics .Furthermore, these models were incorporated into a single web platform with defined pairwise tanimoto similarity-based applicability domain and hosted in the provided link: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://mutli-model-cirrhosis.streamlit.app/\u003c/span\u003e\u003cspan address=\"https://mutli-model-cirrhosis.streamlit.app/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e allowing researchers to utilize the predictive models effectively. Moreover, early identification of potential responders through such computational tools may reduce trial-and-error prescribing, minimize adverse effects, and improve overall treatment outcomes in patients with chronic liver diseases.\u003c/p\u003e","manuscriptTitle":"Smart Simulations for Scarred Livers: Target-Based Computational Models in Cirrhosis Care","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-10 15:35:21","doi":"10.21203/rs.3.rs-7704948/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"bd43bbb7-5039-4f57-8d6f-bf9343b2cc1c","owner":[],"postedDate":"October 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-10-21T10:16:48+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-10 15:35:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7704948","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7704948","identity":"rs-7704948","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}