Recep Adiyaman
bioinformatics

Issue #22: Advantages and Limitations of AlphaFold in Structural Biology: Insights from Recent Studies.

January 14, 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

Advantages and Limitations of AlphaFold in Structural Biology: Insights from Recent Studies.

🧬 Abstract

Over the past three years, AlphaFold-a deep learning-based protein structure prediction system-has transformed structural biology by providing near-experimental accuracy models directly from amino acid sequences. This narrative review synthesizes applications reported in the 2022-2025 literature across human, microbial, and viral systems, drawing on peer-reviewed studies as our data source. Representative examples include modeling of SARS-CoV-2 spike and nucleocapsid proteins in virology, assisting cryo-EM interpretation of bacterial ribosomal and membrane-protein complexes in microbiology, and refining conformational hypotheses for human GPCRs in biomedicine. Across these cases, AlphaFold predictions have complemented experimental workflows by accelerating hypothesis generation, improving model fitting within ambiguous density regions (poorly resolved areas of cryo-EM maps), and guiding mutagenesis strategies to probe dynamic conformational states. We also summarize recent method extensions: AlphaFold-Multimer improves multi-chain complex assembly prediction, while molecular dynamics (MD) simulations augment AlphaFold’s static models by sampling conformational flexibility and testing stability. Despite these advances, important limitations remain-particularly for intrinsically disordered regions, protein-ligand and protein-cofactor interactions, and very large or transient assemblies-and current community benchmarks indicate that approximately one-third of residues may lack atomistic precision, underscoring uncertainty in flexible or modified segments. Framed within a clear chronological window and evidence base, our analysis highlights both the practical impact and the remaining challenges of integrating AlphaFold with experiment, outlining priorities where further methodological innovation and orthogonal validation are needed.

Why it matters: Critical for improving fold accuracy and reducing structural uncertainty in de novo design.

⭐ Additional Signals

Benchmarking co-folding methods to predict the structures of covalent protein-ligand complexes.

Targeted covalent inhibitors (TCIs) are emerging as a new modality in drug discovery because of their strong binding affinity and prolonged target engagement. However, the rational design of TCIs remains a significant challenge and is hindered by the lack of methods that accurately predict the structures of covalent protein-ligand complexes. Recent advances in co-folding approaches have made substantial strides in modeling complex biomolecular structures. Despite significant progress, their performance profiles for predicting the structures of covalent protein-ligand complexes remain largely unexplored because of the absence of rigorous benchmarks. Here, we introduce CoFD-Bench, a comprehensive benchmark dataset comprising 218 recently resolved covalent complexes designed to systematically evaluate both classical docking methods (AutoDock-GPU, CovDock, and GNINA) and deep learning co-folding models (AlphaFold3 (AF3), Chai-1, and Boltz-1x). Our results demonstrate that co-folding methods achieve superior ligand RMSD accuracy and protein-ligand interaction recovery. However, their performance markedly declines for novel pocket-ligand pairs. In contrast, classical docking methods exhibit stable but modest performance, which is primarily limited by target conformations. Furthermore, computational efficiency evaluations show that co-folding methods are slower than classical approaches, posing challenges for large-scale predictions. We also reveal that AF3 has the potential to identify native covalent residues through noncovalent co-folding, with a ligand RMSD comparable to that of covalent co-folding. These findings offer a possible route to explore covalent binding without prior specification of reactive residues, which are often unknown in real-world scenarios. Our study provides crucial insights and new opportunities for future co-folding-based TCI design, informing future model applications and improvements. CoFD-Bench offers rigorous evaluation criteria, diverse docking scenarios, and various methodological baselines, positioning it as an important benchmark for future model development and assessment.

Benchmarking all-atom biomolecular structure prediction with FoldBench.

Accurate prediction of biomolecular complex structures is fundamental for understanding biological processes and rational therapeutic design. Recent advances in deep learning methods, particularly all-atom structure prediction models, have significantly expanded their capabilities to include diverse biomolecular entities, such as proteins, nucleic acids, ligands, and ions. However, comprehensive benchmarks covering multiple interaction types and molecular diversity remain scarce, limiting fair and rigorous assessment of model performance and generalizability. To address this gap, we introduce FoldBench, an extensive benchmark dataset consisting of 1522 biological assemblies categorized into nine distinct prediction tasks. Our evaluations reveal critical performance dependencies, showing that ligand docking accuracy notably diminishes as ligand similarity to the training set decreases, a pattern similarly observed in protein-protein interaction modeling. Furthermore, antibody-antigen predictions remain particularly challenging, with current methods exhibiting failure rates exceeding 50%. Among evaluated models, AlphaFold 3 consistently demonstrates superior accuracy across the majority of tasks. In summary, our results highlight significant advancements yet reveal persistent limitations within the field, providing crucial insights and benchmarks to inform future model development and refinement.

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.

🧪 AI & Research News

🏢 Industry Insight & Applications

⚡ Quick Reads

Mechanisms of Okanin against wound healing based on network pharmacology, molecular docking and molecular dynamics simulation.

Wound healing is a critical aspect of modern medicine, impacting patient health, quality of life, and healthcare resource allocation. Okanin, a flavonoid from the Asteraceae family, has shown potential in promoting wound healing. This study investigates okanin’s key molecular targets, binding affinity, and mechanisms of action using network pharmacology, molecular docking, molecular dynamics simulations, and in vivo experimental validation. Okanin’s potential targets were identified using the Comparative Toxicogenomics Database (CTD) and SwissTargetPrediction, while wound healing-related targets were sourced from GeneCards and DrugBank. Overlap analysis of these datasets revealed common targets. Key target proteins were filtered through protein-protein interaction (PPI) analysis using the STRING database. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were conducted using Metascape to build a drug-target-pathway-disease network. Molecular docking was performed with AutoDockTools, and binding affinity was evaluated through energy scores, particularly with AURKA and HDAC1. Molecular dynamics simulations with GROMACS confirmed the stability of okanin-target complexes. ADME/T properties were assessed using SwissADME and ProTox-3.0 to evaluate pharmacokinetics and toxicity. In vivo quantitative real-time PCR (qRT-PCR) was performed to assess the expression of selected target genes in a mouse wound model following topical okanin treatment. A total of 72 common targets were identified between okanin and wound healing. PPI network analysis highlighted 17 key targets, with molecular docking revealing the highest binding affinity for AURKA and HDAC1 (ΔG = - 8.8 kcal/mol for both). GROMACS were then run on the top complexes. Target-ligand stability was quantified by convergence of RMSD/Rg, sustained hydrogen-bond counts, and MM/GBSA binding free energies (AURKA, - 24.27 ± 3.65 kcal/mol; HDAC1, - 47.7 ± 1.60 kcal/mol), confirming robust interactions. SwissADME predicted good drug-likeness (MW = 288.25 g/mol; logP = 1.69; high GI and moderate skin permeability) and no P-gp liability, while ProTox-3.0 indicated low systemic toxicity (LD₅₀ = 2500 mg/kg). qRT-PCR results demonstrated that okanin treatment significantly downregulated AURKA and PIK3R1, while upregulating HDAC1, in wounded skin, supporting the predicted molecular interactions and regulatory functions. Okanin promotes wound healing through multiple molecular targets and pathways, including antioxidant, anti-inflammatory, and cell proliferation mechanisms. Its high binding affinity for AURKA and HDAC1, along with modulation of the IL-17 and AMPK signaling pathways, underscores its therapeutic potential. This study provides a comprehensive theoretical and experimental framework for the development of okanin as a topical agent for wound healing, with future research focusing on formulation development and translational applications.

Integrative gene target mapping, RNA sequencing, in silico molecular docking, ADMET profiling and molecular dynamics simulation study of marine derived molecules for type 1 diabetes mellitus.

Type 1 diabetes mellitus (T1DM) is a metabolic disease leading threat to human health around the world. Here we aimed to explore new biomarkers and potential therapeutic targets in T1DM through adopting integrated bioinformatics tools. The gene expression Omnibus (GEO) database was used to obtain next generation sequencing data (GSE270484) of T1DM and normal control samples. Furthermore, differentially expressed genes (DEGs) were screened using the DESeq2 package in R bioconductor package. Gene Ontology (GO) and pathway enrichment analyses were performed by g:Profiler. The protein-protein interaction (PPI) network was plotted with IID PPI database and visualized using Cytoscape. Module analysis of the PPI network was done using PEWCC. Then, microRNAs (miRNAs) and transcription factors (TFs) in T1DM were screened out from the miRNet and NetworkAnalyst database. Then, the miRNA-hub gene regulatory network and TF-hub gene regulatory network were constructed by Cytoscape software. Moreover, a drug-hub gene interaction network of the hub genes was constructed and predicted the drug molecule against hub genes. The receiver operating characteristic (ROC) curves were generated to predict diagnostic value of hub genes. Finally we performed molecular docking, ADMET profiling and molecular dynamics simulation studies of marine derived chemical constituents using Schrodinger Suite 2025-1. A total of 958 DEGs were screened: 479 up regulated genes and 479 down regulated genes. DEG were mainly enriched in the terms of developmental process, membrane, cation binding, response to stimulus, cell periphery, ion binding, neuronal system and metabolism. Based on the data of protein-protein interaction (PPI), the top 10 hub genes (5 up regulated and 5 down regulated) were ranked, including FN1, GSN, ADRB2, CEP128, FLNA, CD74, EFEMP2, POU6F2, P4HA2 and BCL6. The miRNA-hub gene regulatory network and TF-hub gene regulatory network showed that hsa-mir-657, hsa-miR-1266-5p, NOTCH1 and GTF3C2 might play an important role in the pathogenesis of T1DM. The drug-hub gene interaction network showed that Clenbuterol, Diethylstilbestrol, Selegiline and Isoflurophate predicted therapeutic drugs for the T1DM. Molecular docking and molecular dynamics simulation study revealed that CMNPD5805 and CMNPD30286 as potential inhibitors of FN1 (pdb id: 3M7P) a key biomarker in pathogenesis of T1DM. These findings promote the understanding of the molecular mechanism and clinically related molecular targets for T1DM.

Structure-Guided Design of Novel Diarylpyrimidine-Based NNRTIs Through a Comprehensive In Silico Approach: 3D-QSAR, ADMET Evaluation, Molecular Docking, and Molecular Dynamics.

Background/Objectives: The emergence of drug-resistant HIV-1 strains challenges the long-term efficacy of current antiretroviral therapies. Non-nucleoside reverse transcriptase inhibitors (NNRTIs) are critical in HIV-1 treatment; however, the need for new candidates with improved resistance profiles and pharmacokinetics remains. This study aims to design and evaluate novel NNRTIs targeting both wild-type (WT) and mutant-type (MT) HIV-1 reverse transcriptase (RT) using integrated computational strategies. Methods: We conducted a 3D-QSAR study on 33 naphthyl-diarylpyrimidine derivatives using CoMFA and CoMSIA models. We designed thirty-five novel molecules based on contour map insights. We applied ADMET and drug-likeness filters to prioritize ten candidates. Molecular docking was performed on WT (PDB: 3HVT) and MT (PDB: 4PUO) RT structures. The top candidates underwent 100 ns molecular dynamics (MD) simulations. We analyzed structural stability via RMSD, RMSF, and Rg, while we used SASA and MolSA to assess solvent exposure and surface compactness. Results: The CoMFA and CoMSIA models demonstrated robust predictivity (R 2 = 0.979/0.920, Q 2 = 0.643/0.546, R 2 test = 0.747/0.603). P14 and P43 showed higher binding affinities than nevirapine and favorable ADMET profiles. MD simulations confirmed stable binding in WT-RT and adaptive flexibility in MT-RT. SASA and MolSA analysis revealed favorable conformational compaction. Drug-likeness profiles indicated optimal log P, strong hydrogen bonding, and acceptable bioavailability. Conclusions: P14 and P43 demonstrate strong potential as NNRTI leads, combining binding affinity, structural stability, and favorable pharmacokinetics, supporting further experimental development.

Potential Mechanisms of Tetramethylpyrazine in the Treatment of Traumatic Brain Injury Based on Network Pharmacology, Molecular Docking, Molecular Dynamics Simulations, and in vivo Experiments.

Background Traumatic brain injury (TBI) is a leading cause of global disability and mortality. Tetramethylpyrazine (TMP), an active compound from Chuanxiong, holds promise for treating cerebrovascular diseases, but its precise mechanism of action against TBI remains incompletely understood. This study aimed to elucidate the therapeutic effects and underlying mechanisms of TMP in TBI. Methods Potential targets of TMP against TBI were identified using Swiss Target Prediction, PharmMapper, and GeneCards databases. Core targets and mechanisms were predicted through network pharmacology, molecular docking, and molecular dynamics (MD) simulations. These computational predictions were then experimentally validated in a rat TBI model, employing behavioral tests, ELISA, RT-qPCR, and Western blot analysis. Results Through network pharmacology analysis, 39 potential targets associated with TMP were identified. Molecular docking and MD simulations manifested that key genes like MMP3, MMP2, MMP13, and GSK3B, showed a strong binding affinity to TMP. GO analysis and KEGG analysis corroborated that such targets strongly related to the IL-17 signaling pathway and the relaxin signaling pathway. In vivo tests proved that TMP could improve the modified Neurological Severity Score (mNSS) and foot defect test scores among rats. ELISA confirmed that TMP could decrease the expression of inflammatory factors, encompassing interleukin 1 beta (IL-1β), interleukin 6 (IL-6), interleukin 17A (IL-17A), and tumor necrosis factor-alpha (TNF-α). Furthermore, RT-qPCR analysis exhibited that the levels of MMP3, MMP2, MMP13, and GSK3B were increased within the rat cortex after TBI. Significantly, TMP treatment alleviated such upregulation. Western blot analysis validated that TMP down-regulated the expression of p-GSK3β (Ser9), active MMP13, active MMP3, and P65 NF-κB proteins after TBI, while TMP increased the expression of occludin protein. Conclusion This study demonstrates that TMP exerts therapeutic effects on TBI by targeting the IL-17 and relaxin signalling pathways, providing evidence for its potential as a clinical therapy.

Molecular Docking Analysis of Some Nrf2 Activators as Therapeutic Agents for the Control of Schistosomiasis

Abstract This study assessed the antischistosomal potentials of Curcumin, resveratrol and sulforaphane in silico . Sequences of Schistosoma mansoni adenylate cyclase, farnesyl diphosphate synthase, geranylgeranyl diphosphate synthase and thioredoxin glutathione reductase were retrieved from UniProt database, then used to query AlphaFold database for structural similarity and the predicted 3D structures were downloaded and saved in PDB format. Curcumin, resveratrol and sulforaphane were searched in PubChem and their 3D structures downloaded in sdf format. The enzymes and ligands were individually imported into PyRx virtual screening tool, underwent universal force field preparation before being converted to the pdbqt format. Molecular docking was then conducted and the interactions visualized using Biovia discovery studio visualizer. The absorption, distribution, metabolism, excretion and toxicity (ADMET) characteristics of the ligands were predicted in ADMETlab3.0. Resveratrol and Curcumin had the highest binding affinities of -8.0 and − 7.4 kcal/mol, respectively with S. mansoni adenylate cyclase. A similar pattern was observed between S. mansoni farnesyl diphosphate synthase with the compounds. Curcumin had the highest binding affinity (-8.5 kcal/mol) with S. mansoni geranylgeranyl diphosphate synthase, followed by resveratrol (–6.9kcal/mol). Binding affinities of -8.0, -7.5, and − 3.3 kcal/mol were exhibited by the complexes of S. mansoni thioredoxin glutathione reductase with curcumin, resveratrol and sulforaphane respectively. The compounds have all passed the Lipinski’s rule of five. Conclusively, curcumin is the most potent inhibitor of S. mansoni thioredoxin glutathione reductase and geranylgeranyl diphosphate synthase, followed by resveratrol, while resveratrol showed most promising activity against S. mansoni adenylate cyclase and farnesyl diphosphate synthase, followed by curcumin.

Synthesis and EGFR binding evaluation of various substituted aurones via molecular docking approaches.

Aurones, belonging to the flavonoid family and widely distributed in various plant species, have attracted significant attention due to their potential inhibitory effects on the epidermal growth factor receptor (EGFR). In this study, a series of substituted aurones was synthesized through the oxidative cyclization of chalcones. Molecular docking studies were conducted using Autodock Vina to evaluate the binding interactions between these aurones and the EGFR kinase domain. The results indicate that aurone derivatives effectively bind to the EGFR active site, with fluorine-substituted aurones (particularly compound 4o) displaying superior binding affinity compared to bromine- and chlorine-substituted analogues. Additionally, methyl-substituted aurones (4a and 4b) exhibited the highest binding affinity of -8.4 kcal/mol. The reliability and stability of the EGFR tyrosine kinase protein model used in the docking studies were validated by the Ramachandran plot analysis. These results support aurone derivatives-especially fluorine- and methyl-substituted scaffolds-as promising EGFR binders with favorable drug-likeness, warranting targeted biochemical validation.

Integrated network toxicology, molecular docking, molecular dynamics simulation, and experimental verification to elucidate the mechanism of hepatotoxicity and processing detoxification in Fructus Meliae Toosendan.

Fructus Meliae Toosendan (FMT) has been traditionally used in Chinese medicine, yet its mechanisms of its hepatotoxicity, as well as the detoxification process following processing, are still not fully elucidateed. This study aimed to investigate the hepatotoxic components and targets from FMT and elucidate both its mechanism of hepatotoxicity and the detoxification mechanism induced by processing. We employed a combination of network pharmacology, molecular docking, molecular dynamics simulations and in vitro experiments to explore the potential hepatotoxic targets and mechnisms of FMT. A total of 12 toxicity components and 71 potential hepatotoxicity targets of FMT were identified. Protein-protein interaction (PPI) network analysis identified the top six core potential targets, three of which possess suitable crystal structures for molecular docking. These include meliasenin B-HSD17B4 (ΔG = -7.40 kcal mol -1 ) and meliasenin B-HMGCR (-7.32 kcal mol -1 ), melianone-HSD17B4 (-8.19 kcal mol -1 ) and melianone-SOD2 (-8.51 kcal mol -1 ). toosendanin-HMGCR (-7.58 kcal mol -1 ). Additionally, 185 Gene Ontology (GO) terms and 72 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were identified. The three major toxicity ingredients and three core targets mentioned above were subjected to molecular docking analysis using AutoDock Vina. Molecular dynamic simulations confirmed a stable interaction between the meliasenin B, melianone and toosendanin and HSD17B4, HMGCR and SOD2 ligand system. It indicated that melianone, toosendanin and isotoosendanin significantly reduced cell viability in a dose-dependent manner in Human hepatoma cells HepG2 and Human normal liver cell line L02 hepatocytes in vitro experiments. Processing FMT enhances viability in both HepG2 and L02 cells while mitigating its hepatotoxic effects. The hepatotoxicity-reducing effect of processing FMT may involve the PI3K/AKT/FOXO signaling pathway. This study suggests that melianone, toosendanin, isotoosendanin and meliasenin B may be hepatotoxic components of FMT, and the detoxification mechanism of processing FMT may be related to the PI3K/AKT/FOXO signaling pathway, warranting further investigation into their safety profiles in clinical applications.

Machine learning-guided repurposing of FDA-approved quinolones as dual cholinesterase inhibitors: A multi-level docking, molecular dynamics, DFT, and SHAP-based analysis.

Alzheimer’s disease (AD) involves progressive cholinergic degeneration, with acetylcholinesterase (AChE) and butyrylcholinesterase (BChE) playing key enzymatic roles in its pathology. In this study, we computationally repurposed four FDA-approved quinolone antibiotics, Norfloxacin, Sparfloxacin, Gatifloxacin, and Nalidixic acid, as potential dual-site cholinesterase (ChE) inhibitors using a hybrid in vitro/in silico workflow. Enzyme inhibition assays identified Norfloxacin as the most potent AChE inhibitor (K I = 1.08 μM), while all compounds displayed non-competitive inhibition toward BChE. Molecular docking and MM-GBSA binding free energy analyses revealed key interactions within the catalytic gorge of AChE, supported by hydrogen bonding with Phe295 and Arg296, as well as π-π contacts with Tyr124. Density functional theory computations highlighted the influence of frontier orbital distribution on binding affinity, particularly for Norfloxacin and Sparfloxacin. An explicit-solvent molecular dynamics simulation of the AChE-Norfloxacin complex further confirmed the stability of the docking-derived binding mode over 100 ns. In an exploratory fashion, SHAP-based machine learning models were applied to a descriptor set derived from QikProp, SwissADME, and Jaguar outputs, suggesting that BBB-related indices and HOMO energy contribute to AChE inhibition, whereas the energy gap is more relevant for BChE; these trends, however, are constrained by the small four-compound dataset and should be regarded as hypothesis-generating. In silico ADME/Tox profiling indicated favorable oral drug-like properties, low predicted CYP450 inhibition liabilities, and physicochemical profiles compatible with CNS-oriented optimization, although passive BBB permeability was not predicted to be high. Finally, systems-level enrichment (STRING, GeneCards) provided a qualitative network context linking ACHE and BCHE to neurodegeneration. Together, these data position Norfloxacin and Sparfloxacin as computationally prioritised candidates whose ChE-related repurposing potential warrants further validation in dedicated cellular and in vivo models.

💡 Pipeline Tip

Pin reference genomes by checksum to avoid version drift.

🛠️ Resources

The protein structure is the language of life; design is its poetry. — Recep Adiyaman

BS HF DK