Pub Date : 2025-12-18DOI: 10.1038/s43588-025-00940-4
Sophia Chen
Data-center operators try to recycle retired hardware, but a broken global recycling infrastructure stands in the way.
数据中心运营商试图回收退役硬件，但全球回收基础设施的缺陷阻碍了这一进程。
{"title":"The afterlife of 20 million AI chips","authors":"Sophia Chen","doi":"10.1038/s43588-025-00940-4","DOIUrl":"10.1038/s43588-025-00940-4","url":null,"abstract":"Data-center operators try to recycle retired hardware, but a broken global recycling infrastructure stands in the way.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"6 1","pages":"2-5"},"PeriodicalIF":18.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145783912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-18DOI: 10.1038/s43588-025-00928-0
Jeremie Alexander, Jonathan M. Stokes
SynGFN integrates synthesis constraints directly into the chemical design process. The result is a generative framework that produces diverse, high-quality molecules that can be readily synthesized in the laboratory.
{"title":"AI-guided molecular design with recipes included","authors":"Jeremie Alexander, Jonathan M. Stokes","doi":"10.1038/s43588-025-00928-0","DOIUrl":"10.1038/s43588-025-00928-0","url":null,"abstract":"SynGFN integrates synthesis constraints directly into the chemical design process. The result is a generative framework that produces diverse, high-quality molecules that can be readily synthesized in the laboratory.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"6 1","pages":"13-14"},"PeriodicalIF":18.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145783857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-16DOI: 10.1038/s43588-025-00936-0
Vishwanathan Akshay, Mile Gu
A recent study demonstrates the applicability of quantum computers for multi-objective optimization, bringing quantum computing a step closer towards practical applications.
最近的一项研究证明了量子计算机对多目标优化的适用性，使量子计算向实际应用更近了一步。
{"title":"Improving the balance of trade-offs in multi-objective optimization with quantum computing","authors":"Vishwanathan Akshay, Mile Gu","doi":"10.1038/s43588-025-00936-0","DOIUrl":"10.1038/s43588-025-00936-0","url":null,"abstract":"A recent study demonstrates the applicability of quantum computers for multi-objective optimization, bringing quantum computing a step closer towards practical applications.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"5 12","pages":"1102-1103"},"PeriodicalIF":18.3,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145770272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-16DOI: 10.1038/s43588-025-00935-1
{"title":"Toward a domain-grounded AI collaborator with SciSciGPT.","authors":"","doi":"10.1038/s43588-025-00935-1","DOIUrl":"https://doi.org/10.1038/s43588-025-00935-1","url":null,"abstract":"","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":" ","pages":""},"PeriodicalIF":18.3,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145770292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-16DOI: 10.1038/s43588-025-00910-w
Zijing Gao, Rui Jiang
Scouter, a deep learning approach, predicts transcriptional responses to genetic perturbations by integrating large language model (LLM)-based gene embeddings with a lightweight compressor–generator neural network, providing valuable insights into the application of LLMs to biological research.
{"title":"Harnessing LLMs to decode genetic perturbations","authors":"Zijing Gao, Rui Jiang","doi":"10.1038/s43588-025-00910-w","DOIUrl":"10.1038/s43588-025-00910-w","url":null,"abstract":"Scouter, a deep learning approach, predicts transcriptional responses to genetic perturbations by integrating large language model (LLM)-based gene embeddings with a lightweight compressor–generator neural network, providing valuable insights into the application of LLMs to biological research.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"6 1","pages":"11-12"},"PeriodicalIF":18.3,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145770173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-15DOI: 10.1038/s43588-025-00918-2
A physics-infused heterogeneous graph neural network has been developed to address challenges in designing complex nanomaterials with spatially varying compositions. This fully differentiable model enables the rapid optimization and discovery of photon upconverting nanoparticle heterostructures that are 6.5-fold brighter than any nanoparticle in the training set.
{"title":"Deep learning accelerates discovery of complex nanomaterials","authors":"","doi":"10.1038/s43588-025-00918-2","DOIUrl":"10.1038/s43588-025-00918-2","url":null,"abstract":"A physics-infused heterogeneous graph neural network has been developed to address challenges in designing complex nanomaterials with spatially varying compositions. This fully differentiable model enables the rapid optimization and discovery of photon upconverting nanoparticle heterostructures that are 6.5-fold brighter than any nanoparticle in the training set.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"6 1","pages":"19-20"},"PeriodicalIF":18.3,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145764732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-10DOI: 10.1038/s43588-025-00922-6
Luca Manneschi, Matthew O. A. Ellis
A recent study demonstrates the efficiency of quantum-mechanical modeling of material properties by mapping the problem onto neuromorphic device architectures.
最近的一项研究通过将问题映射到神经形态器件架构上，证明了材料特性量子力学建模的效率。
{"title":"Predicting physics efficiently with hybrid hardware","authors":"Luca Manneschi, Matthew O. A. Ellis","doi":"10.1038/s43588-025-00922-6","DOIUrl":"10.1038/s43588-025-00922-6","url":null,"abstract":"A recent study demonstrates the efficiency of quantum-mechanical modeling of material properties by mapping the problem onto neuromorphic device architectures.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"5 12","pages":"1104-1105"},"PeriodicalIF":18.3,"publicationDate":"2025-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145727761","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-09DOI: 10.1038/s43588-025-00909-3
Dinghao Wang, Qingrun Zhang
A framework called AUTOENCODIX benchmarks diverse autoencoder architectures in biological molecular profiling data, enabling insights from complex, multi-layered data.
Pub Date : 2025-12-09DOI: 10.1038/s43588-025-00933-3
Bart Ghesquiere
Mass spectrometry data analysis has long been limited to known molecules and exact matches. In a recent manuscript, a scalable search algorithm is proposed for uncovering both known compounds and novel molecular variants, enabling insights into natural product biosynthesis.
{"title":"A scalable tool for fast and flexible variant identification in mass spectrometry","authors":"Bart Ghesquiere","doi":"10.1038/s43588-025-00933-3","DOIUrl":"10.1038/s43588-025-00933-3","url":null,"abstract":"Mass spectrometry data analysis has long been limited to known molecules and exact matches. In a recent manuscript, a scalable search algorithm is proposed for uncovering both known compounds and novel molecular variants, enabling insights into natural product biosynthesis.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"5 12","pages":"1108-1109"},"PeriodicalIF":18.3,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145717052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-09DOI: 10.1038/s43588-025-00916-4
Maximilian Josef Joas, Neringa Jurenaite, Dušan Praščević, Nico Scherf, Jan Ewald
In recent years, autoencoders, a family of deep learning-based methods for representation learning, are advancing data-driven research owing to their variability and nonlinear power for multimodal data integration. Despite their success, current implementations lack standardization, versatility, comparability and generalizability. Here we present AUTOENCODIX, an open-source framework, designed as a standardized and flexible pipeline for preprocessing, training and evaluation of autoencoder architectures. These architectures, such as ontology-based and cross-modal autoencoders, provide key advantages over traditional methods by offering explainability of embeddings or the ability to translate across data modalities. We apply the method to datasets from pan-cancer studies (The Cancer Genome Atlas) and single-cell sequencing as well as in combination with imaging. Our studies provide important user-centric insights and recommendations to navigate through architectures, hyperparameters and important tradeoffs in representation learning. These include the reconstruction capability of input data, the quality of embedding for downstream machine learning models and the reliability of ontology-based embeddings for explainability. An open-source framework called AUTOENCODIX is developed to enable reproducible comparison of vanilla, variational, stacked, ontology-based and cross-modal autoencoders.
{"title":"AUTOENCODIX: a generalized and versatile framework to train and evaluate autoencoders for biological representation learning and beyond","authors":"Maximilian Josef Joas, Neringa Jurenaite, Dušan Praščević, Nico Scherf, Jan Ewald","doi":"10.1038/s43588-025-00916-4","DOIUrl":"10.1038/s43588-025-00916-4","url":null,"abstract":"In recent years, autoencoders, a family of deep learning-based methods for representation learning, are advancing data-driven research owing to their variability and nonlinear power for multimodal data integration. Despite their success, current implementations lack standardization, versatility, comparability and generalizability. Here we present AUTOENCODIX, an open-source framework, designed as a standardized and flexible pipeline for preprocessing, training and evaluation of autoencoder architectures. These architectures, such as ontology-based and cross-modal autoencoders, provide key advantages over traditional methods by offering explainability of embeddings or the ability to translate across data modalities. We apply the method to datasets from pan-cancer studies (The Cancer Genome Atlas) and single-cell sequencing as well as in combination with imaging. Our studies provide important user-centric insights and recommendations to navigate through architectures, hyperparameters and important tradeoffs in representation learning. These include the reconstruction capability of input data, the quality of embedding for downstream machine learning models and the reliability of ontology-based embeddings for explainability. An open-source framework called AUTOENCODIX is developed to enable reproducible comparison of vanilla, variational, stacked, ontology-based and cross-modal autoencoders.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"6 1","pages":"96-108"},"PeriodicalIF":18.3,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s43588-025-00916-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145717100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：