Nicolas Laverde, Christian Grévisse, Sandra Jaramillo, Ruben Manrique
{"title":"Integrating large language model-based agents into a virtual patient chatbot for clinical anamnesis training.","authors":"Nicolas Laverde, Christian Grévisse, Sandra Jaramillo, Ruben Manrique","doi":"10.1016/j.csbj.2025.05.025","DOIUrl":"10.1016/j.csbj.2025.05.025","url":null,"abstract":"<p><p>Effective communication is crucial for trust-building, accurate information gathering, and clinical decision-making in healthcare. Despite its emphasis in medical curricula, traditional training methods, such as role-playing with standardized patients, remain costly, logistically complex, and fail to replicate real-life scenarios. Simulation-based training enhances communication and reasoning skills, but novice learners often struggle due to underdeveloped reasoning processes. Furthermore, limited access to asynchronous, autonomous simulated patient interactions restricts personalized practice. Virtual patient models offer scalable solutions with interactive scenarios and tailored feedback, but high development costs and resource demands hinder their widespread adoption. To address these challenges, virtual patient systems powered by Large Language Models (LLMs) have emerged as a promising tool. These generative agents simulate human-like behavioral responses by leveraging LLM capabilities, cognitive mechanisms, and contextual memory retrieval. A tool was developed allowing students to select clinical cases and interact with a chatbot simulating a patient role. Teachers can also create custom cases. Evaluations showed that the agent provided consistent, plausible responses aligned with case descriptions and achieved a Chatbot Usability Questionnaire (CUQ) score of 86.25/100. Our results show that this approach enables flexible, repetitive, and asynchronous practice while offering real-time feedback.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2481-2491"},"PeriodicalIF":4.4,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12180958/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144474177","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nathan R Johnson, Fabian Gonzalez-Toro, Barbara Bernal Gomez
{"title":"Class-agnostic annotation of small RNAs balances sensitivity and specificity in diverse organisms.","authors":"Nathan R Johnson, Fabian Gonzalez-Toro, Barbara Bernal Gomez","doi":"10.1016/j.csbj.2025.05.045","DOIUrl":"10.1016/j.csbj.2025.05.045","url":null,"abstract":"<p><p>Small RNAs (sRNAs) are important regulatory elements in eukaryotic organisms and comprise the functional elements of RNA interference. Numerous classes of sRNAs have been annotated, however they vary greatly in their ease of annotation and compatibility with most annotators. Significant challenges exist for the annotation process, including variation in sRNA library quality, alignment depth, and poorly defined loci, collectively making this process difficult. Additionally, few annotators are fully agnostic to sRNA classes and may struggle identifying loci in less explored organisms. To address these problems, we present the annotation tool YASMA-tradeoff (YTO), which is specifically suited to finding reliable thresholds for locus annotation which balance sensitivity with specificity. We compared YTO with other annotators, we show that it and other pipelines based on coverage-normalization methods have great advantages, balancing many metrics to produce a more reproducible annotation. We also demonstrate that YTO produces more contiguous and representative loci, through the aggressive merging of similar expressed regions. Finally, we also show that the tool produces much more descriptive locus dimensions, a major advantage in species where sRNAs may be distinct or unique. Overall, we demonstrate substantial improvements in annotation accuracy, reproducibility, and description, particularly in non-model organisms and less-explored clades.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2450-2459"},"PeriodicalIF":4.4,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12174571/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144324615","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Does isovaleric acid play a key role in the interaction between probiotics and antidepressants? A secondary analysis of a randomized clinical trial.","authors":"Oliwia Gawlik-Kotelnicka, Marharyta Sobczak, Joanna Palma, Marta Popławska, Karolina Skonieczna-Żydecka, Maksymilian Plewka, Rafał Pawliczak, Dominik Strzelecki","doi":"10.1016/j.csbj.2025.05.035","DOIUrl":"10.1016/j.csbj.2025.05.035","url":null,"abstract":"<p><strong>Background: </strong>Dysbiosis appears to be a significant contributor to the complex pathophysiology of mood disorders, and short-chain fatty acids (SCFAs), the main metabolites produced in the colon by bacterial fermentation, have been found to play a role in gut-brain communication. Probiotics were shown to be effective in managing and alleviating depressive symptoms, especially as an add-on protocol. This study aimed to assess the change in fecal SCFAs levels after supplementation with probiotics in patients with depression, depending on baseline antidepressant treatment.</p><p><strong>Methods: </strong>This was a secondary analysis of a two-arm, parallel-group, randomized, double-blind, controlled trial. Data from 65 participants were analyzed. The intervention included probiotic formulation (<i>Lactobacillus helveticus</i> Rosell®-52 and <i>Bifidobacterium longum</i> Rosell®-175; R0052/R0175) or placebo over a 60-day period. Then, stratification was performed by the type of antidepressant medications. Fecal SCFAs were measured by the gas chromatography method. Pre-intervention socio-demographic, clinical, and laboratory data were assessed.</p><p><strong>Results: </strong>Probiotics used decreased the levels of isovaleric acid compared with placebo when administered with non-selective serotonin reuptake inhibitors antidepressants (non-SSRIs) with large effect size (p = .019, |r|=.653), but not when used with SSRIs (p = .572, |r|=.109) or applied alone (p = .404, |r|=.182). Isovalerate levels decreased as depression improved in the probiotic plus non-SSRIs group. Conclusions: R0052/R0175 as an add-on to non-SSRI antidepressants may offer antidepressant action partly through the decrease in isovaleric acid levels. More research with a larger sample size is needed to study SCFAs' role as a mediator of antidepressant action of both probiotics and medications. ClinicalTrials.gov identifier: NCT04756544.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2275-2287"},"PeriodicalIF":4.4,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12164015/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144301301","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"MFE-DDI: A multi-view feature encoding framework for drug-drug interaction prediction.","authors":"Lingfeng Wang, Yinghong Li, Yaozheng Zhou, Liping Guo, Congzhou Chen","doi":"10.1016/j.csbj.2025.05.029","DOIUrl":"10.1016/j.csbj.2025.05.029","url":null,"abstract":"<p><p>Multidrug combination therapy has long been a vital approach for treating complex diseases by leveraging synergistic effects between drugs. However, drug-drug interactions (DDIs) are not uniformly beneficial. Accurate and rapid identification of DDIs is critical to mitigate drug-related side effects. Currently, many computational-based methods have been used to expedite the prediction of DDIs. However, most of these methods use a single perspective to obtain drug features, which have limited expressive capabilities and cannot fully represent the essential attributes of drugs. In this study, we propose the Multi-view Feature Embedding for drug-drug interaction prediction (MFE-DDI), which integrates SMILES information, molecular graph data and atom spatial semantic information to model drugs from multiple perspectives and encapsulate the intricate drug information crucial for predicting DDIs. Concurrently, the feature information extracted from different feature encoding channels is fused in the attention-based fusion module to fully convey the essence of drugs. Consequently, this approach enhances the efficacy of the DDI prediction task. Experimental results indicate that MFE-DDI surpasses other baseline methods on three datasets. Moreover, analysis experiments demonstrate the robustness of the model and the necessity of each component of the model. Case studies on newly approved drugs demonstrate the effectiveness of our method in real scenarios. The code and data used in MFE-DDI can be found at https://github.com/2019040445/MFE_DDI.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2473-2480"},"PeriodicalIF":4.4,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12181007/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144474154","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hoang-Anh T Vo, Sang Nguyen, Ai-Quynh T Tran, Han Nguyen, Hai Bich Ho, Philip W Fowler, Timothy M Walker, Thuy Thi Nguyen
{"title":"Deep learning-based framework for Mycobacterium tuberculosis bacterial growth detection for antimicrobial susceptibility testing.","authors":"Hoang-Anh T Vo, Sang Nguyen, Ai-Quynh T Tran, Han Nguyen, Hai Bich Ho, Philip W Fowler, Timothy M Walker, Thuy Thi Nguyen","doi":"10.1016/j.csbj.2025.05.030","DOIUrl":"10.1016/j.csbj.2025.05.030","url":null,"abstract":"<p><p>Tuberculosis (TB) kills more people annually than any other pathogen. Resistance is an ever-increasing global problem, not least because diagnostics remain challenging and access limited. 96-well broth microdilution plates offer one approach to high-throughput phenotypic testing, but they can be challenging to read. Automated Mycobacterial Growth Detection Algorithm (AMyGDA) is a software package that uses image processing techniques to read plates, but struggles with plates that exhibit low growth or images of low quality. We developed a new framework, TMAS (TB Microbial Analysis System), which leverages state-of-the-art deep learning models to detect <i>M. tuberculosis</i> growth in images of 96-well microtiter plates. TMAS is designed to measure Minimum Inhibitory Concentrations (MICs) rapidly and accurately while differentiating between true bacterial growth and artefacts. Using 4,018 plate images from the CRyPTIC (Comprehensive Resistance Prediction for Tuberculosis: An International Consortium) dataset to train models and refine the framework, TMAS achieved an essential agreement of 98.8%, significantly outperformed the 90% threshold established by the International Organization for Standardization (ISO). TMAS offers a reliable, automated and complementary evaluation to support expert interpretation, potentially improving accuracy and efficiency in tuberculosis drug susceptibility testing (DST).</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2208-2218"},"PeriodicalIF":4.4,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12166714/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144301300","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Andrea Riquelme-García, Juan Mulero-Hernández, Jesualdo Tomás Fernández-Breis
{"title":"Annotation of biological samples data to standard ontologies with support from large language models.","authors":"Andrea Riquelme-García, Juan Mulero-Hernández, Jesualdo Tomás Fernández-Breis","doi":"10.1016/j.csbj.2025.05.020","DOIUrl":"10.1016/j.csbj.2025.05.020","url":null,"abstract":"<p><p>The semantic integration of biological data is hindered by the vast heterogeneity of data sources and their limited semantic formalization. A crucial step in this process is mapping data elements to ontological concepts, which typically involves substantial manual effort. Large Language Models (LLMs) have demonstrated potential in automating complex language-related tasks and may offer a solution to streamline biological data annotation. This study investigates the utility of LLMs-specifically various base and fine-tuned GPT models-for the automatic assignment of ontological identifiers to biological sample labels. We evaluated model performance in annotating labels to four widely used ontologies: the Cell Line Ontology (CLO), Cell Ontology (CL), Uber-anatomy Ontology (UBERON), and BRENDA Tissue Ontology (BTO). Our dataset was compiled from publicly available, high-quality databases containing biologically relevant sequence information, which suffers from inconsistent annotation practices, complicating integrative analyses. Model outputs were compared against annotations generated by text2term, a state-of-the-art annotation tool. The fine-tuned GPT model outperformed both the base models and text2term in annotating cell lines and cell types, particularly for the CL and UBERON ontologies, achieving a precision of 47-64% and a recall of 88-97%. In contrast, base models exhibited significantly lower performance. These results suggest that fine-tuned LLMs can accelerate and improve the accuracy of biological data annotation. Nonetheless, our evaluation highlights persistent challenges, including variable precision across ontology categories and the continued need for expert curation to ensure annotation validity.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2155-2167"},"PeriodicalIF":4.4,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12162076/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144282753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Graph convolutional neural networks improved target-specific scoring functions for cGAS and kRAS in virtual screening.","authors":"Bo Wang, Muhammad Junaid, Wenjin Li","doi":"10.1016/j.csbj.2025.05.023","DOIUrl":"10.1016/j.csbj.2025.05.023","url":null,"abstract":"<p><p>In the drug discovery process, traditional structure-based virtual screening methods, such as molecular docking, are often limited by empirical scoring functions. Recent studies have demonstrated that target-specific scoring functions developed using machine learning approaches can enhance the accuracy of virtual screening. Furthermore, the extrapolation performance of these scoring functions is crucial for their broader applicability. Therefore, the study tried combining molecular graph and convolutional neural networks as a way to improve the extrapolation ability of target-specific scoring functions in the face of data expanded within a certain range of chemical space. Taking cGAS and kRAS proteins as examples, through rigorous data screening and feature extraction, the study constructed multiple supervised learning models containing traditional machine learning models, and deep learning models like graph convolutional networks. The results show that compared with the generic scoring functions, these target-specific scoring functions showed significant superiority. In addition, the target-specific scoring functions also exhibit remarkable robustness and accuracy in determining whether a molecule is active. This indicates that the graph convolutional network can be generalized to the prediction of heterogeneous data based on the complex patterns of molecular protein binding that have been learned. The comprehensive performance evaluation of different target-specific scoring functions shows that they hold significant potential for applications in structure-based virtual screening. In particular, graph convolutional networks was demonstrated to greatly improve the screening efficiency and accuracy of target-specific scoring functions for targets such as cGAS and kRAS.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2176-2185"},"PeriodicalIF":4.4,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12162078/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144282754","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jinshil Kim, Chin-Hsien Tai, Natalie K Livingston, Jennifer Patterson-West, Oliver Stearns, Bokyung Son, Deborah M Hinton
{"title":"Tevenvirinae phages encode a family of LSm-like fold proteins.","authors":"Jinshil Kim, Chin-Hsien Tai, Natalie K Livingston, Jennifer Patterson-West, Oliver Stearns, Bokyung Son, Deborah M Hinton","doi":"10.1016/j.csbj.2025.05.028","DOIUrl":"10.1016/j.csbj.2025.05.028","url":null,"abstract":"<p><p>Uncharacterized bacteriophage proteins typically have little homology outside the phage world. An example is the T4 early protein GoF. Although the function of wild type <i>goF</i> is not known, the GoF mutant (D25Y) affects the level of T4 gene <i>41</i> mRNA under certain conditions. To investigate possible GoF functions, we leveraged the power of AlphaFold3. We found that despite having very dissimilar sequences, GoF and 2 other uncharacterized T4 early proteins, MotB.1 and Frd.2, are structurally similar with predicted N-terminal LSm-like fold motifs. Since this motif, which is found throughout biology, is frequently associated with an RNA function and the GoF(D25Y) mutation is found within the predicted LSm-like fold, we hypothesized that these proteins may affect gene expression. Consequently, we used a fluorescent translational <i>mCherry</i> reporter system and RT-qPCR to investigate if and how the presence of the proteins affect the expression of an <i>mCherry</i> gene placed downstream of the T4 gene <i>41</i> 5' untranslated region. We find that the heterologous expression of <i>goF</i>(D25Y) increases the level of mCherry post-transcriptionally by increasing the stability of the RNA. However, neither WT GoF nor MotB.1 have this effect. We speculate that GoF(D25Y) may represent a gain-of-function mutant that can increase RNA stability. Using AlphaFold3 models we speculate how the D25Y mutation in GoF might facilitate or enhance RNA binding. Our work reveals the power of AlphaFold to find unexpected structure/function relationships among uncharacterized proteins.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2051-2062"},"PeriodicalIF":4.4,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12152347/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144274316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Daniel Spitzl, Markus Mergen, Ulrike Bauer, Friederike Jungmann, Keno K Bressem, Felix Busch, Marcus R Makowski, Lisa C Adams, Florian T Gassert
{"title":"Leveraging large language models for accurate classification of liver lesions from MRI reports.","authors":"Daniel Spitzl, Markus Mergen, Ulrike Bauer, Friederike Jungmann, Keno K Bressem, Felix Busch, Marcus R Makowski, Lisa C Adams, Florian T Gassert","doi":"10.1016/j.csbj.2025.05.019","DOIUrl":"10.1016/j.csbj.2025.05.019","url":null,"abstract":"<p><strong>Background & aims: </strong>The rapid advancement of large language models (LLMs) has generated interest in their potential integration in clinical workflows. However, their effectiveness in interpreting complex (imaging) reports remains underexplored and has at times yielded suboptimal results. This study aims to assess the capability of state-of-the-art LLMs to classify liver lesions based solely on textual descriptions from MRI reports, challenging the models to interpret nuanced medical language and diagnostic criteria.</p><p><strong>Methods: </strong>We evaluated multiple LLMs, including GPT-4o, Deepseek V3, Claude 3.5 Sonnet, and Gemini 2.0 Flash, on a physician-generated fictitious dataset of 88 MRI reports designed to resemble real clinical radiology documentation. The dataset included a representative spectrum of common liver lesions, such as hepatocellular carcinoma, cholangiocarcinoma, hemangiomas, metastases, and focal nodular hyperplasia. Model performance was assessed using micro and macro F1-scores benchmarked against ground truth labels.</p><p><strong>Results: </strong>Claude 3.5 Sonnet demonstrated the highest diagnostic accuracy among the evaluated models, achieving a micro F1-score of 0.91, outperforming other LLMs in lesion classification.</p><p><strong>Conclusion: </strong>These findings highlight the feasibility of LLMs for text-based diagnostic support, particularly in resource-limited or high-volume clinical settings. While LLMs show promise in medical diagnostics, further validation through prospective studies is necessary to ensure reliable clinical integration. The study emphasizes the importance of rigorous benchmarking to assess model performance comprehensively.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2139-2146"},"PeriodicalIF":4.4,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12158552/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144274315","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Application of temperature replica exchange molecular dynamics: Structure of mitotic spindle-associated protein SHE1 and its binding to dynein.","authors":"Laleh Alisaraie, Oliver Stueker, Sayi'Mone Tati","doi":"10.1016/j.csbj.2025.05.024","DOIUrl":"10.1016/j.csbj.2025.05.024","url":null,"abstract":"<p><p>Cytoskeletal motor protein dynein belongs to the AAA+ superfamily of enzymes, functioning as a mechanochemical ATPase that converts chemical energy into force to drive its retrograde movements along microtubules. Dynein is responsible for cellular cargo transportation; however, viral particles can also recruit dynein. Dynein's mutation is also critical in neurodegenerative and neurodevelopmental diseases. SHE1 is a yeast-specific MT-associated protein that promotes polarizing dynein-mediated spindle movements. Unlike dynein's adaptor proteins, SHE1 is the only protein known to inhibit dynein motility, act independently from dynactin, and alter dynein activity. Despite SHE1's unique mode of action, its structure has not yet been solved experimentally. This work presents the SHE1 structure obtained using Temperature Replica Exchange Molecular Dynamics simulations. The resulting structure was used to explore the conformations of the complex formed by SHE1 binding to dynein and/or microtubule. The conformations of the complex obtained from the computational protein-protein binding study were clustered using the unsupervised machine learning K-means algorithm. The results helped identify the potential SHE1-dynein interaction sites and the participating amino acids, as well as explaining the structural details underlying SHE1's potential inhibitory mechanisms. In one of the two main recognized binding sites of SHE1 in the SHE1-dynein complexes, its inhibitory mechanism can be due to its interference with the long-range allosteric communications of dynein's domains, namely strut-stalk-MTBD. In that binding mode, SHE1 can restrain the AAA1/AAA4 modules of the motor ring, affecting its \"open-closed\" conformational changes. That suggests SHE1 could directly interfere with the ATP-hydrolyzing modules necessary for dynein motility. In the second observed binding site, SHE1 interacts with MTBD, α-tubulin, and the C-terminal tail of β-tubulin (E-hook) thereby inhibiting the high binding affinity mode of MTBD to microtubules preventing its motility, which aligns with recent <i>in vitro</i> experimental data. Characterizing the SHE1 structure and its complex with SHE1-dynein can aid in the design and development of therapeutic peptide inhibitors of dynein or its mutants for treating dynein-involved diseases.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"2359-2374"},"PeriodicalIF":4.4,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12172990/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144316013","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}