Recep Adiyaman
bioinformatics

Issue #25: Enhancing CYP450-Ligand Binding Predictions: A Comparative Analysis of Ligand-Based and Hybrid Machine Learning Models.

January 19, 2026 Daily Intelligence
Protein Design Daily

Building something in Protein Design?

I love collaborating on new architectural challenges. Let's build together.

LinkedIn Connect Direct Message

🧬 Protein Design Digest

Curated protein signals by Recep Adiyaman

Join 1,000+ researchers. Unsubscribe anytime.

🚀 Today’s Top Signal

Enhancing CYP450-Ligand Binding Predictions: A Comparative Analysis of Ligand-Based and Hybrid Machine Learning Models.

🧬 Abstract

Predicting cytochrome P450 (CYP450) ligand binding is critical in early-stage drug discovery as CYP450-mediated metabolism profoundly influences drug efficacy, safety, and adverse reaction risks. However, experimental determination of CYP450-ligand interactions remains resource- and time-intensive, underscoring the need for robust computational alternatives. While ligand-based methods are commonly employed, they often fail to fully account for structural intricacies governing protein-ligand interactions. To address this gap, we developed a hybrid machine learning framework integrating ligand descriptors, protein descriptors, and protein-ligand interaction descriptors that include molecular docking-derived parameters, rescoring function components from multiple algorithms, and structural interaction fingerprints (SIFt). Evaluated on CYP1A2 and CYP17A1 isoforms, our model demonstrated superior predictive accuracy in cross-validation compared with stand-alone molecular docking and ligand-based approaches. Furthermore, benchmarking against state-of-the-art tools (SwissADME and ADMETlab 3.0) revealed enhanced performance in binding prediction. This work establishes a versatile framework for advancing computational tools to prioritize CYP450 binding assessments during drug discovery.

Why it matters: Essential ground-truth data for validating next-gen foundation models like Boltz or Chai.

⭐ Additional Signals

Predicting the Mechanism of Action of Bawei Chufan Soup in Treating Teen Depression through Network Pharmacology, Molecular Docking and Molecular Dynamics Simulation.

Introduction The Bawei Chufan Soup (BWCFS) in Traditional Chinese Medicine (TCM) offers unique advantages in treating Teen Depression (TD). This study utilizes network pharmacology, molecular docking, and molecular dynamics simulations to predict the material basis and mechanism of action of the decoction. Methods The TCMSP, SwissADME, and SwissTargetPrediction databases were utilized to obtain the active ingredients and targets of the BWCFS. The GeneCards, OMIM, and Disgenet databases were used to identify disease targets, and the intersection of these sets was determined using the VENNY tool. The intersecting targets were imported into the String database for protein- protein interaction analysis and the screening of core targets. GO and KEGG enrichment analyses of the intersecting targets were conducted using the David database, and drugcomponent- target-pathway network diagrams were constructed using Cytoscape 3.10.0 software. The molecular docking models of the core components and key targets were generated using AutoDock Vina, and kinetic simulations were conducted using GROMACS 2020.3, paired with the best docking models. Results After screening, the study identified the core components of BWCFS as Baicalein, Kaempferol, Quercetin, Cerevisterol, and Cavidine, with the key targets for TD being AKT1, IL6, TNF, ESR1, and IL1B. GO enrichment analysis revealed that BWCFS may affect signal transduction in the treatment of TD, and is associated with cellular components such as the plasma membrane and dendrites, as well as the regulation of protein binding. KEGG analysis suggested that the intersecting genes are primarily enriched in the cyclic adenosine monophosphate (cAMP) signaling pathway. Molecular docking results indicated that AKT1 shows good binding affinity with Baicalein, Cavidine, Kaempferol, and Quercetin, while Cerevisterol exhibits strong binding with TNF. The molecular dynamics simulations were stable and reliable. During the protein-ligand complex simulation, the binding between the protein and ligand was stable, with van der Waals interactions as the primary force, while hydrogen bonds were present between both the protein and ligand. Discussion Though this study has several common limitations associated with network pharmacology, and no animal experiments have been conducted for verification, the study has successfully explored and validated the mechanism of action of BWCFS in treating TD using scientific computational methods. This study provides new perspectives and methods for the development and management of pharmacological treatments for TD, offering innovative insights into TCM approaches for its treatment. Conclusion Through network pharmacology, this study preliminarily predicted the material basis and mechanism of action of BWCFS in treating TD. Furthermore, the therapeutic effects of BWCFS on TD may be associated with neuroinflammation and structural and functional changes in neuronal dendrites. The cAMP-PKA-NF-κB and cAMP-PI3K-AKT-NF-κB pathways are proposed as potential therapeutic targets.

MetalloDock: Decoding Metalloprotein-Ligand Interactions via Physics-Aware Deep Learning for Metalloprotein Drug Discovery.

Accurate prediction of metalloprotein-ligand interactions is critical for metalloprotein-targeted drug discovery. Conventional docking tools and existing deep learning (DL) models fail to reliably capture metal-ligand interactions, hampering the discovery of potent metalloprotein inhibitors. Here, we propose MetalloDock, the first DL-based docking framework specially designed for metalloprotein targets. By innovatively integrating an autoregressive spatial decoding engine with a physics-constrained geometric generation paradigm, MetalloDock can precisely reconstruct metal coordination geometries and accurately capture metal-ligand interactions, which enhance both the accuracy of metalloprotein-ligand docking and binding affinity prediction. Extensive evaluations on our custom-built benchmark data set demonstrate that MetalloDock outperforms existing methods, including AlphaFold3, in docking success rate and virtual screening performance for metalloprotein targets. In real-world applications, MetalloDock successfully identified multiple novel hit compounds in a virtual screening campaign targeting the prostate-specific membrane antigen. Additionally, it enabled rational drug design for acidic polymerase endonuclease, leading to the discovery of potent inhibitors. These results highlight the broad applicability of MetalloDock in accelerating metalloprotein-targeted drug discovery and provide a standardized framework for future evaluation of metalloprotein-specific docking algorithms.

In silico screening of IMPPAT-derived phytochemicals targeting ERG6 and drug resistance-associated proteins in drug-resistant Candida albicans: virtual screening and molecular dynamics using alphafold models.

Pathogenic fungi, particularly Candida albicans, have been escalating clinical problems, notably because of antifungal resistance and symptomatological comorbidity with COVID-19. This research aimed to find phytochemical inhibitors of ergosterol production, specifically targeting ERG6 (C-24 sterol methyltransferase), utilizing chemicals from the IMPPAT database. A total of 14,965 phytochemicals were computationally evaluated against AlphaFold-predicted ERG6 utilizing AutoDock Vina. Fifteen compounds exhibiting robust binding affinities (- 8.2 to - 9.2 kcal/mol) were found, from which four candidates were chosen based on advantageous ADMET profiles. The docking scores for the top four compounds targeting ERG6-Daturataturin A (- 8.8 kcal/mol), Disogluside (- 8.6), Tataramide B (- 8.4), and Floribundasaponin A (- 8.4)-exceeded those of previously identified ERG6 inhibitors D28 (- 8.0), Tomatidine (- 7.9), and H55 (- 6.4). The selected leads were further docked against other proteins associated with drug resistance and cell proliferation, specifically ERG1, ERG11, CLB2, CDR1, and CDR2. Among these, only ERG1 exhibited significant interactions, with Disogluside (- 9.3 kcal/mol), Tataramide B (- 9.9), and Floribundasaponin A (- 9.3) surpassing the reference inhibitor terbinafine (- 8.7 kcal/mol), except for Daturataturin A, which showed a comparable score of - 8.6 kcal/mol. Nevertheless, owing to steric conflicts inside the ERG1 binding sites, molecular dynamics (MD) simulations were conducted exclusively for ERG6-ligand complexes over duration of 100 ns. The RMSD values demonstrated commendable structural stability: Daturataturin A (~ 0.39 nm), Disogluside (~ 0.38 nm), Tataramide B (~ 0.27 nm), and Floribundasaponin A (~ 0.40 nm). Principal Component Analysis (PCA) validated consistent and significant movements for Daturataturin A and Floribundasaponin A, whereas Disogluside and Tataramide B exhibited increased flexibility. MM/PBSA analysis indicated robust binding free energies for Daturataturin A (- 42.26 kcal/mol), Floribundasaponin A (- 37.48 kcal/mol), and Disogluside (- 29.58 kcal/mol), however Tataramide B exhibited a detrimental + 9.81 kcal/mol. These results endorse the promise of phytochemical-derived antifungals and necessitate more experimental verification. The online version contains supplementary material available at 10.1007/s40203-025-00480-9.

🧪 AI & Research News

🏢 Industry Insight & Applications

⚡ Quick Reads

RLBindDeep: A ResNet-LSTM based novel framework for protein-ligand binding affinity prediction.

The prediction of the binding affinity of proteins and ligands in computational drug discovery with high accuracy is critical when evaluating the effectiveness of potential therapeutic compounds. This research work introduces RLBindDeep, a novel deep learning architecture based on the amalgamation of the ResNet and LSTM architectures, for improved accuracy in predicting protein-ligand binding affinities. Most traditional methodologies utilizing conventional molecular docking techniques suffer from poor accuracy owing to semi-flexible modeling approaches and limited considerations of complex interactions. On the other hand, RLBindDeep, which is formulated as a pose-independent binding affinity regression model that directly predicts experimental protein-ligand binding affinities from fixed complex structures, without performing docking or rescoring multiple poses, has performed well in extracting important features of the protein-ligand interaction. Specifically, the extracted features encompass ligand physicochemical descriptors (e.g., molecular weight, LogP, TPSA), protein-level features such as amino acid composition, and detailed interaction features including van der Waals, electrostatic, and hydrogen-bond energies. The model has been tested rigorously over the CASF-2016 benchmark dataset and has returned Pearson’s coefficient R=0.875, Spearman’s coefficient ρ=0.864, and Root Mean Square Error RMSE=0.993. This significantly outperforms existing state-of-the-art models, such as HAC-Net and AutoDock Vina. Improved accuracy and robustness in RLBindDeep further highlight the possibility of deep learning to revolutionize computational drug discovery processes, making strategies for drug development more efficient and targeted.

In-Silico Evaluation of Novel Quercetin Derivatives as Antimalarial Agents Targeting Plasmodium falciparum Dihydrofolate Reductase: Molecular Docking, Molecular Dynamics Simulations, and Pharmacokinetic and Toxicity Predictions

Abstract The global rise in malaria mortality is largely driven by the increasing resistance of Plasmodium falciparum to widely used and affordable antimalarial drugs, emphasizing the urgent need for novel therapeutic agents. In this study, thirty-five ligands were computationally evaluated as potential inhibitors of the bifunctional enzyme dihydrofolate reductase–thymidylate synthase (DHFR-TS; PDB ID: 1J3K), a validated antimalarial drug target. Virtual screening and molecular docking were performed using AutoDock implemented in PyRx v0.8, followed by molecular dynamics simulations using GROMACS v2024.4. Pharmacokinetic and toxicity profiles were assessed using SwissADME and Protox 3.0. ADME analysis revealed favorable pharmacokinetic properties for most compounds, while toxicity predictions classified the majority into Classes IV and V, indicating low acute toxicity, with all promising ligands predicted to be non-hepatotoxic, non-cardiotoxic, and non-carcinogenic. Among the evaluated compounds, six ligands (b.II, b.III, b.VIII, I.b, II.b, and VII.b) exhibited higher binding affinities toward DHFR-TS than the reference drugs cycloguanil and pyrimethamine. These ligands formed strong hydrogen bonding and van der Waals interactions with key active-site residues, including SER A:167, LEU A:40, ILE A:14, LEU A:164, ASN A:33, THR A:36, GLU A:25, and CYS A:27, contributing to enhanced binding stability. Notably, ligand b.II demonstrated the strongest binding affinity and sustained stability, identifying it as a promising lead candidate for further experimental validation against multidrug-resistant Plasmodium falciparum.

Cresol Derivatives from Bacillus subtilis as Natural Oviposition Modulator of Culex quinquefasciatus: A Molecular Docking Approach.

Mosquitoes rely heavily on olfactory cues for locating suitable oviposition sites, with microbial communities in aquatic habitats playing a crucial role in producing volatile organic compounds (VOCs) that influence mosquito behaviour. In this study, we isolated Bacillus subtilis DHB13 from the breeding habitat of Culex quinquefasciatus, a major vector of several human diseases. The partial 16S rRNA gene sequence of the isolate has been submitted to NCBI GenBank with the accession number PV698100. The identity and resistance profile of the strain was confirmed through biochemical and antibiotic susceptibility tests. The bacterial suspension demonstrated a notable oviposition activity index (OAI) of 0.77 ± SE, with moderate variation among treatments (F(3, 8) = 3.631, p = 0.0642). Multiple comparison analysis (Tukey’s test) showed that OAI values for DHB13-treated media did not differ significantly from natural habitat water but were significantly higher than the sterile control, indicating a biologically relevant attraction of gravid female mosquitoes. LC-MS analysis of the bacterial culture supernatant revealed the presence of three cresol derivatives: diisopropyl-m-cresol, 3-ethyl-p-cresol, and 6-ethyl-o-cresol. These compounds were evaluated through molecular docking against Cx. quinquefasciatus Odorant Binding Protein 1 (CxOBP1), a protein known to mediate olfactory-driven oviposition behaviour. However, mosquito olfaction involves several OBPs, receptors, and enzymes, so interaction with CxOBP1 represents only part of this complex sensory system. Molecular docking revealed strong binding of CxOBP1 with diisopropyl-m-cresol (-6.7 kcal/mol), 3-ethyl-p-cresol (-6.2 kcal/mol), and 6-ethyl-o-cresol (-5.9 kcal/mol), indicating potential oviposition attractant activity. All three ligands were found to bind within a conserved binding pocket of CxOBP1, behavioural assays confirmed the oviposition-stimulant properties of the bacterial suspension, indicating that the detected compounds mimic natural semio-chemicals such as p-cresol, previously recognized as an oviposition cue. These findings reinforce the role of microbiota in shaping mosquito reproductive behaviour through the production of volatile attractants. Moreover, they highlight the potential of using microbial VOCs as environmentally sustainable tools for mosquito surveillance and vector control. This integrative approach linking microbial ecology, chemical analysis, and mosquito behaviour provides novel insights for the development of attractant-based control strategies.

Unraveling Multi-target Mechanisms of Codonopsis pilosula in Breast Cancer: A Synergistic Approach Combining Network Pharmacology, Molecular Docking, and Machine Learning Techniques.

Introduction Breast cancer is a leading cause of cancer-related mortality in women. Although the traditional Chinese medicine Codonopsis Pilosula (CP) is empirically used in its treatment, the underlying mechanisms of action remain elusive. This study aimed to apply a novel integrative network pharmacology and machine learning approach to identify bioactive compounds in CP and elucidate their anti-breast cancer mechanisms. Methods The analysis utilized a comprehensive and innovative workflow that combined network pharmacology, machine learning-based target prediction, bioinformatics analyses, and molecular docking and molecular dynamics simulations. Publicly available datasets were mined for CP constituents and putative targets, and integrated with breast cancer-associated gene profiles. Key compound-target interactions were prioritized via machine learning algorithms. Results Machine learning highlighted EGFR and PTGS2 as primary targets. Molecular docking and dynamics demonstrated stable binding of Taraxerol and Stigmasterol to these proteins, with EGFR-Taraxerol, EGFR-Spinasterol, PTGS2-Stigmasterol, and PTGS2-Taraxerol complexes exhibiting robust affinity and stability. Discussion The findings are significant as they reveal previously unreported interactions between CP’s bioactive compounds and critical breast cancer targets. This provides a molecularlevel explanation for the traditional use of CP, bridging the gap between TCM and modern pharmacology. These results offer a solid foundation for further experimental validation. Conclusion This multidisciplinary, predictive strategy successfully identified key bioactive compounds in CP and their molecular targets in breast cancer. The study provides crucial mechanistic evidence for CP’s therapeutic potential and highlights the power of this integrated approach for drug discovery from TCM (Traditional Chinese Medicine).

Action mechanism of Qianlie Xiaozheng decoction against prostate cancer: network pharmacology, molecular docking, and molecular dynamics simulations.

Prostate cancer (PCa) is a leading male malignancy. This study explores the anti-PCa mechanism of Qianlie Xiaozheng decoction (QLXZD) using network pharmacology. From 34 ingredients and 23 potential therapeutic targets, 3 hub ingredients (baicalein, kaempferol, quercetin) and 4 hub targets (CCNB1, CDK1, EGFR, TOP2A) were prioritized. Enrichment analysis of the 23 targets linked them to cell cycle and kinase signaling. Molecular docking confirmed strong binding of the hub ingredients to the hub targets, comparable to known inhibitors. Molecular dynamics simulations supported baicalein-TOP2A complex stability. These findings reveal QLXZD exerts anti-PCa effects via a multi-component, multi-target mechanism, supporting its clinical application.

AI-guided integrative discovery of PD-1/PD-L1 interface inhibitors through multiscale modeling and experimental validation.

The PD-1/PD-L1 protein-protein interaction (PPI) is a critical immune checkpoint, and its inhibition represents a powerful strategy in oncology. Disrupting this macromolecular complex with small molecules remains a significant challenge. This study establishes a comprehensive pipeline for the discovery of novel in stock PD-1/PD-L1 inhibitors. We first developed a robust machine learning-based quantitative structure-activity relationship (ML-QSAR) model to screen chemical libraries virtually. Top-ranking hits were subjected to molecular docking against the PD-L1 dimer interface to evaluate potential binding modes. Subsequently, extensive molecular dynamics (MD) simulations provided critical insights into the structural stability and dynamic interactions at the macromolecular interface, revealing the compounds’ mechanism of complex disruption. The most promising candidate, designated PDA13, was advanced to in vitro validation, demonstrating direct binding to the PD-L1 protein and effectively inhibiting the PD-1/PD-L1 interaction with an IC 50 of hit 17.53 μM. Our work underscores the synergy of computational and experimental strategies in targeting PPI of PD-L1 dimer. The identified PDA13 scaffold provides a valuable starting point for the development of novel immunotherapeutic agents, and the detailed biophysical and structural insights into its mechanism of action form the core of this contribution.

Machine learning-guided discovery of mitogen-activated protein kinase 7 (MAPK7 inhibitors): integrating virtual screening, docking, and molecular dynamics simulations.

Cancer remains a major global health challenge and is the second leading cause of mortality worldwide. Despite extensive efforts, the development of effective cancer therapies is still limited. Mitogen-activated protein kinase 7 (MAPK7), a critical regulator of cell proliferation, gene transcription, and metabolism, has recently emerged as a promising therapeutic target for cancer intervention. In this study, we applied advanced machine learning-based computational approaches to identify potential MAPK7 inhibitors. Virtual screening of a large library of drug-like molecules using machine learning models identified 33 active compounds against MAPK7. Molecular docking further refined these hits to five compounds with favorable binding affinities and strong interactions with key catalytic residues. Molecular dynamics (MD) simulations provided additional insights into the stability and conformational dynamics of protein-ligand complexes, highlighting amino acid residues crucial for inhibitor retention within the active site. Collectively, our findings suggest that these five compounds represent promising MAPK7 inhibitors, offering new opportunities for the development of targeted cancer therapeutics. To the best of our knowledge, this is the first study to combine machine learning-based virtual screening, molecular docking, and MD simulations for the identification of MAPK7 inhibitors. Supplementary information The online version contains supplementary material available at 10.1007/s40203-025-00531-1.

Mechanistic study of plastic monomers in gestational diabetes mellitus: A network toxicology and molecular docking approach.

Plastics are widely used in various fields such as food packaging, textile fibers, building materials, and transportation. Although the relationship between plastic additives and diseases has been reported, there is limited research on the association between plastic monomers (PM) and gestational diabetes mellitus (GDM). This study aims to investigate the link between environmental PM and GDM. By employing advanced network toxicology and molecular docking techniques, we successfully elucidated the molecular mechanisms by which PM may induce GDM. Utilizing databases such as PubChem, SEA, Super-PRED, SwissTargetPrediction, PharmMapper, Gene Cards, and OMIM, we identified potential targets associated with the disease. Further analysis using STRING and Cytoscape software helped determine the core targets most significantly related to these metabolic disorders. Additionally, Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analyses were conducted using the David database to characterize these core targets. Finally, molecular docking with CB-Dock2 was used to validate the binding affinity of PM to these target proteins. Our findings suggest that PM may potentially induce GDM by modulating the insulin signaling pathway through STAT3, AKT1, and TP53. In summary, this work provides novel insights into the mechanisms by which environmental pollutants may trigger GDM, thereby laying a theoretical foundation for disease prevention and treatment. It offers valuable references for the safety evaluation of plastics, urging food safety regulatory agencies to strengthen oversight and encouraging the public to reduce plastic usage.

💡 Pipeline Tip

Verify FASTA headers for special characters that break Rosetta pipelines.

🛠️ Resources

Deep learning is not a magic wand, but a powerful lens for structural biology. — Recep Adiyaman

BS HF DK