Computational and structural biotechnology journal最新文献

ChatGPT-4V prompt: A tool to enhance junior radiologists' diagnostic capabilities in cystic renal masses to senior-level accuracy ChatGPT-4V提示：将初级放射科医生对囊性肾肿块的诊断能力提高到高级水平的工具

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.030

Dong-dong Jin , Nan Zhang , Yan Wang , Ke Lin , Bin Qiao , Yue Yang , Jin-hua Lin , Ding-xiang Xie , Xiao-yan Xie , Xiao-hua Xie , Bo-wen Zhuang

Purpose

This study aimed to evaluate the diagnostic performance of prompt-adapted ChatGPT-4 Vision (ChatGPT-4V) in interpreting contrast-enhanced ultrasound (CEUS) images for the assessment of cystic renal masses (CRMs) using Bosniak classification. Additionally, it tested the ability of the best prompt to assist radiologists with different experience.

Materials and methods

This retrospective study included 103 CRMs in patients who underwent CEUS and CT. ChatGPT-4V and six radiologists (three senior and three junior) independently assigned the Bosniak category (BC) based solely on CEUS images. Subsequently, radiologists re-assessed these images while reviewing ChatGPT-4V-generated BC and decided whether to modify their initial assessments. The diagnostic performance of radiologists and prompts was assessed using the area under the receiver operating characteristic curve (AUC).

Results

The AUCs for prompts ranged from 0.507 to 0.688, whereas that for radiologists ranged from 0.685 to 0.831. Among all prompts, Reflection of Thoughts (ROT) prompt achieved the highest AUC, demonstrating performance comparable to juniors (0.688 vs. 0.715, P = 0.727). Although the AUC was lower than that of seniors (0.688 vs. 0.832, P = 0.019), ROT improved the AUCs of juniors: from 0.714 to 0.834 for junior 1, from 0.685 to 0.782 for junior 2, and from 0.704 to 0.783 for junior 3, with the post-assistance performance of all three being comparable to that of seniors.

Conclusion

Prompt-adapted ChatGPT-4V showed variable performance in interpreting CEUS images. ROT as the best-performing prompt achieved diagnostic performance comparable to juniors, and it could help juniors achieve an AUC comparable to that of seniors.

目的：本研究旨在评价ChatGPT-4 Vision （ChatGPT-4V）在解释对比增强超声（CEUS）图像中使用Bosniak分类评估囊性肾肿块（CRMs）的诊断性能。此外，它还测试了最佳提示协助具有不同经验的放射科医生的能力。材料和方法本回顾性研究纳入103例行超声造影和CT检查的crm患者。ChatGPT-4V和6名放射科医生（3名高级和3名初级）仅根据超声造影图像独立分配波斯尼亚类别（BC）。随后，放射科医生在审查chatgpt - 4v生成的BC时重新评估这些图像，并决定是否修改他们的初始评估。放射科医生和提示者的诊断表现是用接受者工作特征曲线（AUC）下的面积来评估的。结果提示医师的auc范围为0.507 ~ 0.688，放射科医师的auc范围为0.685 ~ 0.831。在所有提示中，反思思想（ROT）提示的AUC最高，表现出与初级提示相当的性能（0.688 vs. 0.715, P = 0.727）。虽然AUC低于学长组（0.688 vs. 0.832, P = 0.019），但ROT提高了学长组的AUC：学长组1从0.714提高到0.834，学长组2从0.685提高到0.782，学长组3从0.704提高到0.783，三人的协助后表现与学长组相当。结论自适应的ChatGPT-4V对超声造影图像的解释表现不一。ROT是表现最好的提示符，其诊断性能与大三学生相当，并且可以帮助大三学生达到与大四学生相当的AUC。

{"title":"ChatGPT-4V prompt: A tool to enhance junior radiologists' diagnostic capabilities in cystic renal masses to senior-level accuracy","authors":"Dong-dong Jin , Nan Zhang , Yan Wang , Ke Lin , Bin Qiao , Yue Yang , Jin-hua Lin , Ding-xiang Xie , Xiao-yan Xie , Xiao-hua Xie , Bo-wen Zhuang","doi":"10.1016/j.csbj.2025.12.030","DOIUrl":"10.1016/j.csbj.2025.12.030","url":null,"abstract":"<div><h3>Purpose</h3><div>This study aimed to evaluate the diagnostic performance of prompt-adapted ChatGPT-4 Vision (ChatGPT-4V) in interpreting contrast-enhanced ultrasound (CEUS) images for the assessment of cystic renal masses (CRMs) using Bosniak classification. Additionally, it tested the ability of the best prompt to assist radiologists with different experience.</div></div><div><h3>Materials and methods</h3><div>This retrospective study included 103 CRMs in patients who underwent CEUS and CT. ChatGPT-4V and six radiologists (three senior and three junior) independently assigned the Bosniak category (BC) based solely on CEUS images. Subsequently, radiologists re-assessed these images while reviewing ChatGPT-4V-generated BC and decided whether to modify their initial assessments. The diagnostic performance of radiologists and prompts was assessed using the area under the receiver operating characteristic curve (AUC).</div></div><div><h3>Results</h3><div>The AUCs for prompts ranged from 0.507 to 0.688, whereas that for radiologists ranged from 0.685 to 0.831. Among all prompts, Reflection of Thoughts (ROT) prompt achieved the highest AUC, demonstrating performance comparable to juniors (0.688 vs. 0.715, P = 0.727). Although the AUC was lower than that of seniors (0.688 vs. 0.832, P = 0.019), ROT improved the AUCs of juniors: from 0.714 to 0.834 for junior 1, from 0.685 to 0.782 for junior 2, and from 0.704 to 0.783 for junior 3, with the post-assistance performance of all three being comparable to that of seniors.</div></div><div><h3>Conclusion</h3><div>Prompt-adapted ChatGPT-4V showed variable performance in interpreting CEUS images. ROT as the best-performing prompt achieved diagnostic performance comparable to juniors, and it could help juniors achieve an AUC comparable to that of seniors.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"32 ","pages":"Pages 34-47"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

In silico interactome analysis reveals distinct and complementary metabolic roles of bacteria in stingless bee larval food 硅相互作用组分析揭示了细菌在无刺蜜蜂幼虫食物中的独特和互补的代谢作用

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2026.01.004

Natanael Borges de Avila , Ana Carolina Costa Santos , Joberth Lee Correa , Ana Maria Bonetti , Carlos Ueira-Vieira , Anderson Rodrigues dos Santos

Background

Observational studies in vinegar fermentation suggested a temporal succession between Bacillus cereus and Acetilactobacillus jinshanensis. Here, we reinterpret this pattern in terms of facilitation and niche construction rather than classical competitive succession. We test this ecological model in a distinct biological context — stingless bee larval food — by combining organism-specific interactomes (GenPPI), hub topology (BriCe outliers), and genome-scale metabolic modeling.

Results

The B. cereus interactome shows a diversified hub architecture enriched in environmental sensing, transport, stress response, and secreted effectors—consistent with a pioneer that probes and actively modifies a fresh niche. In contrast, the A. jinshanensis interactome is dominated by a cytoplasm-centric, translation-heavy super-hub—consistent with a fast-growth specialist optimized for efficient biomass production once resources are simplified and available. Flux analyses indicate capacities for xenobiotic processing and terpenoid precursor synthesis in both organisms, but the topological signatures and pathway enrichments align with complementary roles: niche construction and conditioning by B. cereus, followed by rapid exploitation by A. jinshanensis.

conclusion

Our findings support a facilitation-based ecological mechanism: B. cereus functions as an early colonizer and niche engineer—sensing, detoxifying, and depolymerizing complex substrates via secreted enzymes—while A. jinshanensis is a fast-growth specialist whose dominance emerges in the conditioned niche. This reframing resolves the apparent contradiction between a “fast-growth” profile and late-stage dominance and provides mechanistic, systems-level support for complementary roles across distinct environments.

背景对食醋发酵的观察研究表明蜡样芽孢杆菌和金山醋酸乳杆菌之间存在时间演替关系。在这里，我们从促进和生态位构建的角度重新解释了这种模式，而不是传统的竞争继承。我们通过结合生物特异性相互作用组（GenPPI）、枢纽拓扑（BriCe异常值）和基因组尺度代谢模型，在一个独特的生物学背景下测试了这个生态模型——无刺蜜蜂幼虫的食物。结果蜡样芽孢杆菌相互作用组具有丰富的环境感知、运输、应激反应和分泌效应因子的多元化枢纽结构，与探索和主动改变新生态位的先驱一致。相比之下，金山拟南草相互作用组由一个以细胞质为中心、重翻译的超级中心主导，与一个快速生长的专家一致，一旦资源简化和可用，就会优化有效的生物质生产。通量分析表明，这两种生物都具有外源加工和萜类前体合成的能力，但拓扑特征和途径富集程度与互补作用一致：蜡样芽孢杆菌构建和调节生态位，随后是金山芽孢杆菌的快速利用。结论我们的研究结果支持了一个基于促进的生态机制：蜡样芽孢杆菌作为一个早期的定植者和生态位工程师，通过分泌酶来感知、解毒和解聚复杂的底物，而金山芽孢杆菌是一个快速生长的专家，在条件生态位中表现出优势。这种重构解决了“快速增长”概要和后期主导地位之间的明显矛盾，并为不同环境中的互补角色提供了机制的、系统级的支持。

{"title":"In silico interactome analysis reveals distinct and complementary metabolic roles of bacteria in stingless bee larval food","authors":"Natanael Borges de Avila , Ana Carolina Costa Santos , Joberth Lee Correa , Ana Maria Bonetti , Carlos Ueira-Vieira , Anderson Rodrigues dos Santos","doi":"10.1016/j.csbj.2026.01.004","DOIUrl":"10.1016/j.csbj.2026.01.004","url":null,"abstract":"<div><h3>Background</h3><div>Observational studies in vinegar fermentation suggested a temporal succession between <em>Bacillus cereus</em> and <em>Acetilactobacillus jinshanensis</em>. Here, we reinterpret this pattern in terms of facilitation and niche construction rather than classical competitive succession. We test this ecological model in a distinct biological context — stingless bee larval food — by combining organism-specific interactomes (GenPPI), hub topology (BriCe outliers), and genome-scale metabolic modeling.</div></div><div><h3>Results</h3><div>The <em>B. cereus</em> interactome shows a diversified hub architecture enriched in environmental sensing, transport, stress response, and secreted effectors—consistent with a pioneer that probes and actively modifies a fresh niche. In contrast, the <em>A. jinshanensis</em> interactome is dominated by a cytoplasm-centric, translation-heavy super-hub—consistent with a fast-growth specialist optimized for efficient biomass production once resources are simplified and available. Flux analyses indicate capacities for xenobiotic processing and terpenoid precursor synthesis in both organisms, but the topological signatures and pathway enrichments align with complementary roles: niche construction and conditioning by <em>B. cereus</em>, followed by rapid exploitation by <em>A. jinshanensis</em>.</div></div><div><h3>conclusion</h3><div>Our findings support a facilitation-based ecological mechanism: <em>B. cereus</em> functions as an early colonizer and niche engineer—sensing, detoxifying, and depolymerizing complex substrates via secreted enzymes—while <em>A. jinshanensis</em> is a fast-growth specialist whose dominance emerges in the conditioned niche. This reframing resolves the apparent contradiction between a “fast-growth” profile and late-stage dominance and provides mechanistic, systems-level support for complementary roles across distinct environments.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 377-388"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973109","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CRUSADE project: Recycling technologies for ELV components to create a sustainable source of market grade materials for EU applications CRUSADE项目：ELV组件的回收技术，为欧盟应用创造可持续的市场级材料来源

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.016

Iakovos Yakoumis , Anastasia-Maria Moschovi , Olga Thoda , Asimina Katsiapi , Panagiota Stamatopoulou , Alexandros Charitos , Lesia Sandig-Predzymirska , Wambui Wamunyu , Sebeh Adjapong , Walter Murru , Francesco Veglio , Svetlana Zueva , Marco Passadoro , Antreas Afantitis , Dimitris Mintis , Andreas Tsoumanis , Eleni Papadopoulou , George Paloumpis , Nikolaos Theodoropoulos , Vasiliki Zogali , Thomas Abo Atia

The European Union’s ambitions for climate neutrality and digital leadership rely on secure access to CRMs, essential for green and digital technologies but supply risks and geopolitical challenges. The CRUSADE project develops a sustainable, integrated recycling approach targeting underutilized waste streams from ELVs, including PCBs, catalytic converters, batteries and automotive fuel cells. CRUSADE proposes a universal hydrometallurgical process treating up to 500 tonnes of waste annually and recovering approximately 40 tonnes of CRMs at commercial purity. At the core of the process is a MWAL technology that enables a 15–80 % reduction in CO₂ emissions compared to conventional leaching methods, depending on feedstock composition. The process integrates AI-based automation for sorting and treatment, advanced downstream recovery and material traceability via digital tools, aligning with Industry 4.0 principles. Currently, the project is progressing toward scaling up these technologies to a modular pilot unit at TRL7, targeting 15 % faster processing time and 15 % lower product cost relative to current practices. CRUSADE aims to close material loops within the automotive and micromobility sectors, contributing to Europe’s circular economy objectives, raw material independence and climate mitigation goals under the European Green Deal.

欧盟实现气候中和和数字领导地位的雄心有赖于对crm的安全访问，这对绿色和数字技术至关重要，但也存在供应风险和地缘政治挑战。CRUSADE项目开发了一种可持续的综合回收方法，针对低碳汽车未充分利用的废物流，包括多氯联苯、催化转换器、电池和汽车燃料电池。CRUSADE提出了一种通用湿法冶金工艺，每年可处理多达500吨废物，并以商业纯度回收约40吨crm。该工艺的核心是MWAL技术，与传统的浸出方法相比，根据原料成分的不同，该技术可使二氧化碳排放量减少15-80 %。该流程通过数字工具集成了基于人工智能的自动化分拣和处理、先进的下游回收和材料可追溯性，符合工业4.0原则。目前，该项目正朝着将这些技术扩展到TRL7的模块化试验单元的方向发展，目标是将处理时间提高15% %，将产品成本降低15% %。CRUSADE旨在关闭汽车和微型交通部门的材料循环，为欧洲绿色协议下的欧洲循环经济目标、原材料独立性和气候缓解目标做出贡献。

{"title":"CRUSADE project: Recycling technologies for ELV components to create a sustainable source of market grade materials for EU applications","authors":"Iakovos Yakoumis , Anastasia-Maria Moschovi , Olga Thoda , Asimina Katsiapi , Panagiota Stamatopoulou , Alexandros Charitos , Lesia Sandig-Predzymirska , Wambui Wamunyu , Sebeh Adjapong , Walter Murru , Francesco Veglio , Svetlana Zueva , Marco Passadoro , Antreas Afantitis , Dimitris Mintis , Andreas Tsoumanis , Eleni Papadopoulou , George Paloumpis , Nikolaos Theodoropoulos , Vasiliki Zogali , Thomas Abo Atia","doi":"10.1016/j.csbj.2025.12.016","DOIUrl":"10.1016/j.csbj.2025.12.016","url":null,"abstract":"<div><div>The European Union’s ambitions for climate neutrality and digital leadership rely on secure access to CRMs, essential for green and digital technologies but supply risks and geopolitical challenges. The CRUSADE project develops a sustainable, integrated recycling approach targeting underutilized waste streams from ELVs, including PCBs, catalytic converters, batteries and automotive fuel cells. CRUSADE proposes a universal hydrometallurgical process treating up to 500 tonnes of waste annually and recovering approximately 40 tonnes of CRMs at commercial purity. At the core of the process is a MWAL technology that enables a 15–80 % reduction in CO₂ emissions compared to conventional leaching methods, depending on feedstock composition. The process integrates AI-based automation for sorting and treatment, advanced downstream recovery and material traceability via digital tools, aligning with Industry 4.0 principles. Currently, the project is progressing toward scaling up these technologies to a modular pilot unit at TRL7, targeting 15 % faster processing time and 15 % lower product cost relative to current practices. CRUSADE aims to close material loops within the automotive and micromobility sectors, contributing to Europe’s circular economy objectives, raw material independence and climate mitigation goals under the European Green Deal.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"33 ","pages":"Pages 28-42"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973110","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Computational and experimental insights into the interaction of the seaweed-derived steroidal metabolite 11α-hydroxyprogesterone with the glucocorticoid receptor 海藻衍生的甾体代谢物11α-羟孕酮与糖皮质激素受体相互作用的计算和实验见解

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.028

Supatchar Sermsakulwat , Phuphiphat Jaikaew , Tiwtawat Napiroon , Worawat Surarit , Thrissawan Traijitt , Theppanya Charoenrat , Bodee Nutho , Arachaporn Thong-olran , Supenya Chittapun

Seaweed-derived metabolites offer a rich source of bioactive compounds with therapeutic potential. In this study, 112 steroidal metabolites from Sargassum polycystum, Gracilaria fisheri, and Caulerpa lentillifera were computationally screened to identify candidates interacting with the glucocorticoid receptor (GR), a key molecule regulating inflammatory signaling. Among them, 11α-hydroxyprogesterone (SW052) from S. polycystum exhibited the favorable predicted GR binding affinity (−11.6 kcal/mol), comparable to hydrocortisone and medrysone. Molecular docking and 500 ns molecular dynamics simulations revealed a stable GR-SW052 complex stabilized by hydrophobic and van der Waals interactions with MET560, LEU563, LEU566, MET601, MET604, LEU608, LEU732, TYR735, and CYS736. Free energy analysis (MM/PBSA) supported favorable thermodynamic binding, and in silico ADMET evaluation predicted good oral absorption and low toxicity. In vitro assays showed that both SW052 and seaweed lipophilic extract were non-cytotoxic (>70 % cell viability) and significantly inhibited nitric oxide (NO) production in lipopolysaccharide-stimulated RAW 264.7 macrophages (IC₅₀ = 144.05 ± 8.06 and 108.24 ± 4.64 µg/mL, respectively). This study establishes an integrated computational-experimental framework for prioritizing seaweed-derived steroidal metabolites targeting the GR. Using this framework, SW052 was identified as a potential natural compound with stable predicted GR engagement, favorable in silico pharmacokinetic properties, and NO suppression, providing a structure-guided basis for further mechanism and in vivo validation.

海藻衍生的代谢物提供了具有治疗潜力的生物活性化合物的丰富来源。本研究从马尾藻（Sargassum polycystum）、江蓠（Gracilaria fisheri）和Caulerpa lentillifera中筛选了112种甾体代谢物，以确定与糖皮质激素受体（GR）相互作用的候选物，GR是调节炎症信号的关键分子。其中，多糖中的11α-羟孕酮（SW052）表现出较好的预测GR结合亲和力（- 11.6 kcal/mol），与氢化可的松和麦地松相当。分子对接和500 ns分子动力学模拟表明，GR-SW052配合物通过与MET560、LEU563、LEU566、MET601、MET604、LEU608、LEU732、TYR735和CYS736的疏水和van der Waals相互作用稳定。自由能分析（MM/PBSA）支持良好的热力学结合，硅ADMET评价预测良好的口服吸收和低毒性。体外实验表明，SW052和海藻亲脂提取物均无细胞毒性（>；70 %细胞存活率），并显著抑制脂多糖刺激的RAW 264.7巨噬细胞中一氧化氮（NO）的产生（IC50分别为144.05 ± 8.06和108.24 ± 4.64 µg/mL）。本研究建立了一个综合的计算-实验框架，以确定海藻来源的类固醇代谢物针对GR的优先级。在这个框架下，SW052被确定为一种潜在的天然化合物，具有稳定的预测GR作用，有利的硅药代动力学性质和NO抑制，为进一步的机制和体内验证提供了结构指导基础。

{"title":"Computational and experimental insights into the interaction of the seaweed-derived steroidal metabolite 11α-hydroxyprogesterone with the glucocorticoid receptor","authors":"Supatchar Sermsakulwat , Phuphiphat Jaikaew , Tiwtawat Napiroon , Worawat Surarit , Thrissawan Traijitt , Theppanya Charoenrat , Bodee Nutho , Arachaporn Thong-olran , Supenya Chittapun","doi":"10.1016/j.csbj.2025.12.028","DOIUrl":"10.1016/j.csbj.2025.12.028","url":null,"abstract":"<div><div>Seaweed-derived metabolites offer a rich source of bioactive compounds with therapeutic potential. In this study, 112 steroidal metabolites from <em>Sargassum polycystum</em>, <em>Gracilaria fisheri</em>, and <em>Caulerpa lentillifera</em> were computationally screened to identify candidates interacting with the glucocorticoid receptor (GR), a key molecule regulating inflammatory signaling. Among them, 11α-hydroxyprogesterone (SW052) from <em>S. polycystum</em> exhibited the favorable predicted GR binding affinity (−11.6 kcal/mol), comparable to hydrocortisone and medrysone. Molecular docking and 500 ns molecular dynamics simulations revealed a stable GR-SW052 complex stabilized by hydrophobic and van der Waals interactions with MET560, LEU563, LEU566, MET601, MET604, LEU608, LEU732, TYR735, and CYS736. Free energy analysis (MM/PBSA) supported favorable thermodynamic binding, and <em>in silico</em> ADMET evaluation predicted good oral absorption and low toxicity. <em>In vitro</em> assays showed that both SW052 and seaweed lipophilic extract were non-cytotoxic (>70 % cell viability) and significantly inhibited nitric oxide (NO) production in lipopolysaccharide-stimulated RAW 264.7 macrophages (IC<sub>50</sub> = 144.05 ± 8.06 and 108.24 ± 4.64 µg/mL, respectively). This study establishes an integrated computational-experimental framework for prioritizing seaweed-derived steroidal metabolites targeting the GR. Using this framework, SW052 was identified as a potential natural compound with stable predicted GR engagement, favorable <em>in silico</em> pharmacokinetic properties, and NO suppression, providing a structure-guided basis for further mechanism and <em>in vivo</em> validation.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 202-220"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145881147","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Machine learning for drug-target interaction prediction: A comprehensive review of models, challenges, and computational strategies 药物-靶标相互作用预测的机器学习：模型、挑战和计算策略的全面回顾

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.033

Bilal Ahmad , Khmaies Ouahada , Habib Hamam

Drug discovery is a long, resource-intensive process with high failure rates. Traditional experimental identification of drug–target interactions (DTIs) is especially time-consuming and costly. This comprehensive review examines how Artificial Intelligence (AI) and Machine Learning (ML) are transforming DTI prediction, offering substantial potential to reduce drug development time and costs. The review provides a detailed examination of AI/ML-based techniques, detailing data representations for drugs, targets, and their interactions through joint drug–target processing. The review extensively discusses feature extraction and engineering methods, including the construction of interaction-specific features. It also encompasses a broad range of learning paradigms and model architectures, including supervised learning, advanced Graph Neural Networks (GNNs), Deep Learning (DL) models, and hybrid approaches. We further examine training protocols and robust evaluation metrics crucial for assessing model generalization. Ultimately, this review highlights the capacity of these advanced AI/ML methods to deliver more accurate, scalable, and interpretable solutions for DTI prediction. This is crucial for accelerating key stages of drug development, including lead compound identification, off-target profiling, drug repurposing, polypharmacology analysis, and the realization of precision medicine.

药物发现是一个漫长的、资源密集的过程，失败率很高。传统的药物-靶标相互作用（DTIs）的实验鉴定尤其耗时和昂贵。这篇全面的综述探讨了人工智能（AI）和机器学习（ML）如何改变DTI预测，为减少药物开发时间和成本提供了巨大的潜力。本文详细介绍了基于AI/ ml的技术，详细介绍了药物、靶标及其通过联合药物-靶标处理的相互作用的数据表示。本文广泛讨论了特征提取和工程方法，包括构建特定于交互的特征。它还涵盖了广泛的学习范式和模型架构，包括监督学习、高级图神经网络（gnn）、深度学习（DL）模型和混合方法。我们进一步研究了对评估模型泛化至关重要的训练方案和鲁棒评估指标。最后，本综述强调了这些先进的AI/ML方法为DTI预测提供更准确、可扩展和可解释的解决方案的能力。这对于加速药物开发的关键阶段至关重要，包括先导化合物鉴定、脱靶分析、药物再利用、多药理学分析和精准医学的实现。

{"title":"Machine learning for drug-target interaction prediction: A comprehensive review of models, challenges, and computational strategies","authors":"Bilal Ahmad , Khmaies Ouahada , Habib Hamam","doi":"10.1016/j.csbj.2025.12.033","DOIUrl":"10.1016/j.csbj.2025.12.033","url":null,"abstract":"<div><div>Drug discovery is a long, resource-intensive process with high failure rates. Traditional experimental identification of drug–target interactions (DTIs) is especially time-consuming and costly. This comprehensive review examines how Artificial Intelligence (AI) and Machine Learning (ML) are transforming DTI prediction, offering substantial potential to reduce drug development time and costs. The review provides a detailed examination of AI/ML-based techniques, detailing data representations for drugs, targets, and their interactions through joint drug–target processing. The review extensively discusses feature extraction and engineering methods, including the construction of interaction-specific features. It also encompasses a broad range of learning paradigms and model architectures, including supervised learning, advanced Graph Neural Networks (GNNs), Deep Learning (DL) models, and hybrid approaches. We further examine training protocols and robust evaluation metrics crucial for assessing model generalization. Ultimately, this review highlights the capacity of these advanced AI/ML methods to deliver more accurate, scalable, and interpretable solutions for DTI prediction. This is crucial for accelerating key stages of drug development, including lead compound identification, off-target profiling, drug repurposing, polypharmacology analysis, and the realization of precision medicine.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 316-345"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A cooperative learning framework for the integration of metabolomic data from multiple cohorts and common phenotype identification 一个合作学习框架，用于整合来自多个队列和共同表型鉴定的代谢组学数据

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.020

E. Salanon , E. Jules , B. Comte , J. Boccard , E. Pujos-Guillot

Integrating metabolomic data from multiple studies/cohorts could be an efficient strategy to enhance statistical power and identify robust biomarkers. However, challenges associated with batch effects, study-specific biases, and dataset heterogeneity, hinder results reproducibility and translation. Limitations have been reported in shared variable mode data integration approaches, both for early fusion that struggles with inter-study variability, and late fusion that may overlook inter-dataset relationships. Here, we propose a novel cooperative learning framework for metabolomics data integration from multiple studies, designed to improve candidate biomarker discovery by balancing advantages of early and late fusion, while mitigating study-specific confounders. The proposed approach consists in leveraging univariate and multivariate analysis and an optimized loss function. To implement the approach, early-stage integration was based on a multiblock method (MINT-PLS-DA), while separate PLS-DA was used in late fusion. Univariate analysis was performed via a mixed model. The approach was first evaluated in controlled conditions using synthetic data, and then applied to two existing untargeted metabolomics human datasets. Preliminary assessment focused on batch effect reduction across datasets, and agreement between early and late fusion outputs. Using real word data, the results showed that 10% of the initial features were stable across early and late fusion. This showed improved consistency compared to when they were published separately on the integrated dataset. All results demonstrate the ability of the proposed approach to capture the common part of phenotypes. The developed integration model based on cooperative learning leverages the complementary strengths of early and late fusion, offering an efficient solution for metabolomics data integration, enhancing the reliability of potential biomarker discovery.

整合来自多个研究/队列的代谢组学数据可能是提高统计能力和识别稳健生物标志物的有效策略。然而，与批效应、研究特定偏差和数据集异质性相关的挑战阻碍了结果的可重复性和翻译。据报道，共享变量模式数据集成方法存在局限性，包括早期融合与研究间可变性的斗争，以及可能忽略数据集间关系的后期融合。在这里，我们提出了一个新的合作学习框架，用于从多个研究中整合代谢组学数据，旨在通过平衡早期和晚期融合的优势来改善候选生物标志物的发现，同时减轻研究特异性混杂因素。所提出的方法包括利用单变量和多变量分析以及优化的损失函数。为了实现该方法，早期融合基于多块方法（MINT-PLS-DA），而后期融合使用单独的PLS-DA。通过混合模型进行单因素分析。该方法首先在受控条件下使用合成数据进行评估，然后应用于两个现有的非靶向代谢组学人类数据集。初步评估侧重于跨数据集的批量效应减少，以及早期和晚期融合输出之间的一致性。使用真实单词数据，结果表明10%的初始特征在融合的早期和后期是稳定的。这表明，与在集成数据集上单独发布它们相比，一致性得到了改善。所有结果都证明了所提出的方法能够捕获表型的共同部分。开发的基于合作学习的整合模型利用了早期和晚期融合的互补优势，为代谢组学数据整合提供了有效的解决方案，提高了潜在生物标志物发现的可靠性。

{"title":"A cooperative learning framework for the integration of metabolomic data from multiple cohorts and common phenotype identification","authors":"E. Salanon , E. Jules , B. Comte , J. Boccard , E. Pujos-Guillot","doi":"10.1016/j.csbj.2025.12.020","DOIUrl":"10.1016/j.csbj.2025.12.020","url":null,"abstract":"<div><div>Integrating metabolomic data from multiple studies/cohorts could be an efficient strategy to enhance statistical power and identify robust biomarkers. However, challenges associated with batch effects, study-specific biases, and dataset heterogeneity, hinder results reproducibility and translation. Limitations have been reported in shared variable mode data integration approaches, both for early fusion that struggles with inter-study variability, and late fusion that may overlook inter-dataset relationships. Here, we propose a novel cooperative learning framework for metabolomics data integration from multiple studies, designed to improve candidate biomarker discovery by balancing advantages of early and late fusion, while mitigating study-specific confounders. The proposed approach consists in leveraging univariate and multivariate analysis and an optimized loss function. To implement the approach, early-stage integration was based on a multiblock method (MINT-PLS-DA), while separate PLS-DA was used in late fusion. Univariate analysis was performed <em>via</em> a mixed model. The approach was first evaluated in controlled conditions using synthetic data, and then applied to two existing untargeted metabolomics human datasets. Preliminary assessment focused on batch effect reduction across datasets, and agreement between early and late fusion outputs. Using real word data, the results showed that 10% of the initial features were stable across early and late fusion. This showed improved consistency compared to when they were published separately on the integrated dataset. All results demonstrate the ability of the proposed approach to capture the common part of phenotypes. The developed integration model based on cooperative learning leverages the complementary strengths of early and late fusion, offering an efficient solution for metabolomics data integration, enhancing the reliability of potential biomarker discovery.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 346-354"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921375","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SubNExT: Towards accurate, efficient and robust gene expression classification for breast cancer subtyping SubNExT：朝着准确，高效和稳健的乳腺癌亚型基因表达分类

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.027

Karl Paygambar, Roude Jean-Marie, Mallek Mziou-Sallami, Vincent Meyer

Optimizing genomic-based molecular subtyping is key to promote personalized medicine. While neural networks face setbacks regarding tabular data modeling, deep learning has undergone groundbreaking advances across multiple domains, catalyzing further breakthroughs across AI applications. New neural network architectures exhibit enhanced performance, efficiency, and robustness, which could benefit the genomic use-case.

In this study we introduce SubNExT, an optimized shallow CNN with a ConvNeXt backbone using t-SNE and DeepInsight 2D-converted gene expression for breast cancer subtyping. It was compared with other modelization strategies for gene expression data, by optimizing a Transformer, an MLP and XGBoost for unconverted values, a 1D CNN (NeXt-TDNN) for ordered values, and a ViT as an alternative for 2D-converted expression. During evaluation, SubNExT obtains an accuracy of 87.12%, matching the state-of-the-art XGBoost and its 87.24% acc at the top of the benchmark. SubNExT manages this performance with just 76k parameters and the shortest training time, as well as the best stability and robustness among all considered approaches.

By providing accurate, efficient and robust molecular subtyping of breast cancer using gene expression data, SubNExT and its design principles catalyze deep learning adoption in oncogenomics.

优化基于基因组的分子分型是推进个体化医疗的关键。虽然神经网络在表格数据建模方面遭遇挫折，但深度学习在多个领域取得了突破性进展，催化了人工智能应用的进一步突破。新的神经网络架构表现出增强的性能、效率和鲁棒性，这可能有利于基因组用例。在这项研究中，我们引入了SubNExT，这是一个优化的浅CNN，具有ConvNeXt主干，使用t-SNE和DeepInsight 2d转换基因表达用于乳腺癌亚型。通过优化Transformer、MLP和XGBoost的未转换值，优化1D CNN （NeXt-TDNN）的有序值，以及ViT作为2d转换表达的替代方法，对基因表达数据的其他建模策略进行了比较。在评估过程中，SubNExT获得了87.12%的准确率，与最先进的XGBoost及其87.24%的acc在基准测试中的最高水平相匹配。SubNExT仅用76k个参数和最短的训练时间就实现了这种性能，并且在所有考虑的方法中具有最佳的稳定性和鲁棒性。通过使用基因表达数据提供准确、高效和稳健的乳腺癌分子分型，SubNExT及其设计原则促进了深度学习在肿瘤基因组学中的应用。

{"title":"SubNExT: Towards accurate, efficient and robust gene expression classification for breast cancer subtyping","authors":"Karl Paygambar, Roude Jean-Marie, Mallek Mziou-Sallami, Vincent Meyer","doi":"10.1016/j.csbj.2025.12.027","DOIUrl":"10.1016/j.csbj.2025.12.027","url":null,"abstract":"<div><div>Optimizing genomic-based molecular subtyping is key to promote personalized medicine. While neural networks face setbacks regarding tabular data modeling, deep learning has undergone groundbreaking advances across multiple domains, catalyzing further breakthroughs across AI applications. New neural network architectures exhibit enhanced performance, efficiency, and robustness, which could benefit the genomic use-case.</div><div>In this study we introduce SubNExT, an optimized shallow CNN with a ConvNeXt backbone using t-SNE and DeepInsight 2D-converted gene expression for breast cancer subtyping. It was compared with other modelization strategies for gene expression data, by optimizing a Transformer, an MLP and XGBoost for unconverted values, a 1D CNN (NeXt-TDNN) for ordered values, and a ViT as an alternative for 2D-converted expression. During evaluation, SubNExT obtains an accuracy of 87.12%, matching the state-of-the-art XGBoost and its 87.24% acc at the top of the benchmark. SubNExT manages this performance with just 76k parameters and the shortest training time, as well as the best stability and robustness among all considered approaches.</div><div>By providing accurate, efficient and robust molecular subtyping of breast cancer using gene expression data, SubNExT and its design principles catalyze deep learning adoption in oncogenomics.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 412-421"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973104","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

seneR: An R package for comprehensive senescence assessment and its application in type 2 diabetes and osteoarthritis seneR：一个综合衰老评估的R包及其在2型糖尿病和骨关节炎中的应用

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.031

Yi Zhang , Xinming Zhang , Cheng Chen , Bing Li , Yao Lu , Xin Ma , Yunfeng Yang

Background

Cellular senescence is a key driver of aging and chronic diseases. However, accurately identifying senescent cells is challenging due to limitations of conventional biomarkers and senescence heterogeneity. Transcriptome-wide analyses offer powerful tools for deciphering cellular states. Yet, there is a critical gap in computational frameworks for senescence assessment from transcriptomic data.

Methods

We developed the seneR package, which includes functions such as calculating senescence identity scores (SID scores), assessing senescence-related phenotypes, and plotting senescence trajectories, and provides an interactive Shiny interface. We applied seneR to transcriptome datasets from human islets and chondrocytes to investigate the role of senescence in Type 2 Diabetes (T2D) and osteoarthritis (OA). Additionally, in vitro validation confirmed phentolamine (PM)'s potential to delay chondrocyte senescence.

Results

seneR accurately identified senescent cells and revealed senescence-related phenotypes in transcriptome datasets. In T2D, SID scores were significantly higher in elderly islets. Senescent islet cells exhibited diminished responsiveness to nutrient stimuli, linking senescence to impaired insulin secretion. In OA, seneR identified SLPI as a molecule strongly associated with chondrocyte senescence, with PM treatment reducing SID scores. Trajectory analysis revealed chondrocyte senescence progression and potential therapeutic targets. In vitro experiments, PM reversed both IL-1β- and H₂O₂-induced chondrocyte senescence.

Conclusion

Our study demonstrates that seneR is a valuable tool for assessing cellular senescence from transcriptomic data, revealing key phenotypes and potential therapeutic targets in T2D and OA. The identification of SLPI as a senescence-associated molecule and the therapeutic potential of PM highlights the utility of our approach in understanding senescence-related diseases.

细胞衰老是衰老和慢性疾病的关键驱动因素。然而，由于传统生物标志物和衰老异质性的限制，准确识别衰老细胞具有挑战性。转录组分析为破译细胞状态提供了强大的工具。然而，在从转录组学数据评估衰老的计算框架中存在一个关键的差距。方法我们开发了seneR软件包，该软件包包括计算衰老身份评分（SID评分）、评估衰老相关表型和绘制衰老轨迹等功能，并提供了一个交互式的Shiny界面。我们将seneR应用于人类胰岛和软骨细胞的转录组数据集，以研究衰老在2型糖尿病（T2D）和骨关节炎（OA）中的作用。此外，体外验证证实酚妥拉明（PM）延缓软骨细胞衰老的潜力。结果sener能够准确识别衰老细胞，并在转录组数据集中揭示衰老相关表型。在T2D中，老年胰岛的SID评分明显较高。衰老的胰岛细胞表现出对营养刺激的反应减弱，将衰老与胰岛素分泌受损联系起来。在OA中，seneR发现SLPI是与软骨细胞衰老密切相关的分子，PM治疗可降低SID评分。轨迹分析揭示了软骨细胞衰老的进展和潜在的治疗靶点。体外实验表明，PM可逆转IL-1β-和H₂O₂诱导的软骨细胞衰老。我们的研究表明，seneR是一种有价值的工具，可以从转录组学数据来评估细胞衰老，揭示T2D和OA的关键表型和潜在的治疗靶点。SLPI作为衰老相关分子的鉴定和PM的治疗潜力突出了我们的方法在理解衰老相关疾病方面的实用性。

{"title":"seneR: An R package for comprehensive senescence assessment and its application in type 2 diabetes and osteoarthritis","authors":"Yi Zhang , Xinming Zhang , Cheng Chen , Bing Li , Yao Lu , Xin Ma , Yunfeng Yang","doi":"10.1016/j.csbj.2025.12.031","DOIUrl":"10.1016/j.csbj.2025.12.031","url":null,"abstract":"<div><h3>Background</h3><div>Cellular senescence is a key driver of aging and chronic diseases. However, accurately identifying senescent cells is challenging due to limitations of conventional biomarkers and senescence heterogeneity. Transcriptome-wide analyses offer powerful tools for deciphering cellular states. Yet, there is a critical gap in computational frameworks for senescence assessment from transcriptomic data.</div></div><div><h3>Methods</h3><div>We developed the seneR package, which includes functions such as calculating senescence identity scores (SID scores), assessing senescence-related phenotypes, and plotting senescence trajectories, and provides an interactive Shiny interface. We applied seneR to transcriptome datasets from human islets and chondrocytes to investigate the role of senescence in Type 2 Diabetes (T2D) and osteoarthritis (OA). Additionally, in vitro validation confirmed phentolamine (PM)'s potential to delay chondrocyte senescence.</div></div><div><h3>Results</h3><div>seneR accurately identified senescent cells and revealed senescence-related phenotypes in transcriptome datasets. In T2D, SID scores were significantly higher in elderly islets. Senescent islet cells exhibited diminished responsiveness to nutrient stimuli, linking senescence to impaired insulin secretion. In OA, seneR identified SLPI as a molecule strongly associated with chondrocyte senescence, with PM treatment reducing SID scores. Trajectory analysis revealed chondrocyte senescence progression and potential therapeutic targets. In vitro experiments, PM reversed both IL-1β- and H₂O₂-induced chondrocyte senescence.</div></div><div><h3>Conclusion</h3><div>Our study demonstrates that seneR is a valuable tool for assessing cellular senescence from transcriptomic data, revealing key phenotypes and potential therapeutic targets in T2D and OA. The identification of SLPI as a senescence-associated molecule and the therapeutic potential of PM highlights the utility of our approach in understanding senescence-related diseases.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 192-201"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145881146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The potential of deep learning on the discovery of new genes implicated in differences of sex development 深度学习在发现与性别发育差异有关的新基因方面的潜力

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.019

Isabel von der Decken , Hamid Azimi , Anna Lauber-Biason

Despite advances in understanding genetic causes of DSD (differences of sex development), the molecular cause remains unknown for over half of affected individuals. Next-generation sequencing (NGS) has improved diagnosis, but interpreting results can be challenging, especially when no known DSD gene mutations are found, or only variants of unknown significance appear. Identifying new genes involved in sex development from whole exome sequencing (WES) alone is difficult. To overcome this, we introduce “GONAD-ResNet,” a residual convolutional neural network designed to predict novel DSD-associated genes by learning complex patterns in time-dependent single-cell gene expression data. When applied to WES data from six patients (three XX, three XY) with DSD, GONAD-ResNet prioritized genes with expression profiles similar to known DSD genes while disregarding ubiquitous or irrelevant genes. This narrowed the list of potential candidates from around 1000 to a few promising novel genes per patient. This innovative approach accelerates the discovery of new DSD-related genes, opening new research avenues and potentially improving patient outcomes.

尽管在了解DSD（性别发育差异）的遗传原因方面取得了进展，但超过一半的受影响个体的分子原因仍然未知。下一代测序（NGS）改善了诊断，但解释结果可能具有挑战性，特别是当没有发现已知的DSD基因突变，或者只有未知意义的变异出现时。仅通过全外显子组测序（WES）鉴定与性发育有关的新基因是困难的。为了克服这个问题，我们引入了“GONAD-ResNet”，这是一种残差卷积神经网络，旨在通过学习时间依赖性单细胞基因表达数据中的复杂模式来预测新的dsd相关基因。当应用于6例DSD患者（3例XX， 3例XY）的WES数据时，GONAD-ResNet优先考虑与已知DSD基因表达谱相似的基因，而忽略了普遍存在或不相关的基因。这样一来，每位患者的潜在候选基因从1000个左右缩小到几个有希望的新基因。这种创新的方法加速了新的dsd相关基因的发现，开辟了新的研究途径，并有可能改善患者的治疗效果。

{"title":"The potential of deep learning on the discovery of new genes implicated in differences of sex development","authors":"Isabel von der Decken , Hamid Azimi , Anna Lauber-Biason","doi":"10.1016/j.csbj.2025.12.019","DOIUrl":"10.1016/j.csbj.2025.12.019","url":null,"abstract":"<div><div>Despite advances in understanding genetic causes of DSD (differences of sex development), the molecular cause remains unknown for over half of affected individuals. Next-generation sequencing (NGS) has improved diagnosis, but interpreting results can be challenging, especially when no known DSD gene mutations are found, or only variants of unknown significance appear. Identifying new genes involved in sex development from whole exome sequencing (WES) alone is difficult. To overcome this, we introduce “GONAD-ResNet,” a residual convolutional neural network designed to predict novel DSD-associated genes by learning complex patterns in time-dependent single-cell gene expression data. When applied to WES data from six patients (three XX, three XY) with DSD, GONAD-ResNet prioritized genes with expression profiles similar to known DSD genes while disregarding ubiquitous or irrelevant genes. This narrowed the list of potential candidates from around 1000 to a few promising novel genes per patient. This innovative approach accelerates the discovery of new DSD-related genes, opening new research avenues and potentially improving patient outcomes.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 221-234"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhanced MiRISC expression noise reduction by self-feedback regulation of mRNA degradation 通过自我反馈调节mRNA降解增强MiRISC表达降噪

IF 4.1 2区生物学 Q2 BIOCHEMISTRY & MOLECULAR BIOLOGY

Computational and structural biotechnology journal

Pub Date : 2026-01-01 DOI: 10.1016/j.csbj.2025.12.025

Shuangmei Tian , Ziyu Zhao , Meharie G. Kassie , Fangyuan Zhang , Beibei Ren , Degeng Wang

The microRNA (miRNA) induced silencing complex (miRISC) is the targeting apparatus and arguably the rate-limiting step of the miRNA-mediated regulatory subsystem – a major noise reducing, though metabolically costly, mechanism. Recently, we reported that miRISC channels miRNA-mediated regulatory activity back onto its own mRNAs to form negative self-feedback loops, a noise-reduction technique in engineering and synthetic/systems biology. In this paper, our mathematical modeling predicts that mRNA expression noise exhibits a negative correlation with the degradation rate (K_deg) and is attenuated by self-feedback control of degradation. We also calculated K_deg and expression noise of mRNAs detected in a total-RNA single-cell RNA-seq (scRNA-seq) dataset. As predicted, miRNA-targeted mRNAs exhibited higher K_deg values accompanied by reduced cell-to-cell expression noise, confirming the operational trade-off between noise suppression and the increased metabolic/energetic costs associated with producing these mRNAs subjected to accelerated degradation and translational inhibition. Moreover, consistent with the K_deg self-feedback control model, miRISC mRNAs (AGO1/2/3 and TNRC6A/B/C) exhibited further reduced expression noise. In summary, mathematical-modeling and total-RNA scRNA-seq data-analyses provide evidence that negative self-feedback regulation of mRNA degradation reinforces miRISC, the core machinery of the miRNA-mediated noise-reduction subsystem. To our knowledge, this is the first study to concurrently analyze mRNA degradation dynamics and expression noise, and to demonstrate noise reduction by direct self-feedback regulation of mRNA degradation.

microRNA （miRNA）诱导沉默复合体（miRISC）是miRNA介导的调控子系统的靶向装置，可以说是限速步骤，这是一种主要的降噪机制，尽管代谢成本很高。最近，我们报道了miRISC将mirna介导的调控活动引导回其自身的mrna上，形成负反馈回路，这是工程和合成/系统生物学中的一种降噪技术。在本文中，我们的数学模型预测mRNA表达噪声与降解率（Kdeg）呈负相关，并通过降解的自反馈控制而减弱。我们还计算了在全rna单细胞RNA-seq （scRNA-seq）数据集中检测到的mrna的Kdeg和表达噪声。正如预测的那样，mirna靶向mrna表现出更高的Kdeg值，同时细胞间表达噪声降低，这证实了噪声抑制与产生这些mrna的代谢/能量成本增加之间的操作权衡，这些mrna受到加速降解和翻译抑制。此外，与Kdeg自反馈控制模型一致，miRISC mrna （AGO1/2/3和TNRC6A/B/C）的表达噪声进一步降低。总之，数学建模和全rna scRNA-seq数据分析提供了证据，表明mRNA降解的负自我反馈调节强化了miRISC，这是mirna介导的降噪子系统的核心机制。据我们所知，这是第一个同时分析mRNA降解动力学和表达噪声，并证明通过mRNA降解的直接自反馈调节来降低噪声的研究。

{"title":"Enhanced MiRISC expression noise reduction by self-feedback regulation of mRNA degradation","authors":"Shuangmei Tian , Ziyu Zhao , Meharie G. Kassie , Fangyuan Zhang , Beibei Ren , Degeng Wang","doi":"10.1016/j.csbj.2025.12.025","DOIUrl":"10.1016/j.csbj.2025.12.025","url":null,"abstract":"<div><div>The microRNA (miRNA) induced silencing complex (miRISC) is the targeting apparatus and arguably the rate-limiting step of the miRNA-mediated regulatory subsystem – a major noise reducing, though metabolically costly, mechanism. Recently, we reported that miRISC channels miRNA-mediated regulatory activity back onto its own mRNAs to form negative self-feedback loops, a noise-reduction technique in engineering and synthetic/systems biology. In this paper, our mathematical modeling predicts that mRNA expression noise exhibits a negative correlation with the degradation rate (K<sub>deg</sub>) and is attenuated by self-feedback control of degradation. We also calculated K<sub>deg</sub> and expression noise of mRNAs detected in a total-RNA single-cell RNA-seq (scRNA-seq) dataset. As predicted, miRNA-targeted mRNAs exhibited higher K<sub>deg</sub> values accompanied by reduced cell-to-cell expression noise, confirming the operational trade-off between noise suppression and the increased metabolic/energetic costs associated with producing these mRNAs subjected to accelerated degradation and translational inhibition. Moreover, consistent with the K<sub>deg</sub> self-feedback control model, miRISC mRNAs (AGO1/2/3 and TNRC6A/B/C) exhibited further reduced expression noise. In summary, mathematical-modeling and total-RNA scRNA-seq data-analyses provide evidence that negative self-feedback regulation of mRNA degradation reinforces miRISC, the core machinery of the miRNA-mediated noise-reduction subsystem. To our knowledge, this is the first study to concurrently analyze mRNA degradation dynamics and expression noise, and to demonstrate noise reduction by direct self-feedback regulation of mRNA degradation.</div></div>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"31 ","pages":"Pages 179-191"},"PeriodicalIF":4.1,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145880700","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0