Publications

Newly published peer-reviewed research across journals and publishers, with the original publication date and what was actually tested.

Thu, Oct 15Foundation model Protein design
Upgraded diffusion model scaffolds intricate enzyme active sites from scratch
University of Washington researchers have introduced RFdiffusion3, a generative model that uses free-atom diffusion to build custom protein and RNA structures around precise geometric constraints. Tested across functional enzyme challenges, the system designed synthetic phosphotriesterases that coordinate two metal ions and zinc-dependent proteases tailored to cut transthyretin, a protein implicated in amyloid diseases. Both enzyme campaigns yielded functional catalysts without iterative rounds of laboratory optimization, producing active hits within 96 tested candidates.
Jasper Butcher, Rohith Krishna, Raktim Mitra, and 30 othersJasper Butcher, Rohith Krishna, Raktim Mitra, Seth M. Woodbury, Donghyo Kim, Rafael I. Brent, Yanjing Li, Nathaniel Corley, Andrew Favor, Anqi Chen, Paul T. Kim, Jonathan Funk, Gabriela Bastidas Reinoso, Kejia Wu, Aiko Muraishi, Helen E. Eisenach, Saman Salike, Simon Mathis, Tuscan Rock Thompson, Antonia Ruder, Enisha Sehgal, Andrew Kubaney, Jie Chen, Yuliya Politanska, Brian Coventry, Odin Zhang, Florence J. Hardy, Bo Qiang, Kieran Didi, Maxwell Kazman, Lars L. Schaaf, Frank DiMaio, David Baker fewer authors
University of Washington; The Novo Nordisk Foundation Center for Biosustainability; Technical University of Denmark; and 5 othersUniversity of Washington; The Novo Nordisk Foundation Center for Biosustainability; Technical University of Denmark; Cambridge; ETH Zurich; Oxford; Howard Hughes Medical Institute; NVIDIA fewer institutions
Science
Thu, Oct 15AI drug discovery
Language models pull standardized rare-disease symptoms directly from patient notes
Researchers built RARE-PHENIX, an automated framework that reads unstructured clinical text, standardizes symptoms into formal Human Phenotype Ontology terms, and ranks them by diagnostic importance. The pipeline was trained on records from 2,671 patients across 11 clinical sites in the Undiagnosed Diseases Network, then tested on 16,357 notes from 143 patients at Vanderbilt University Medical Center. The extracted phenotype lists matched clinician curation closely and preserved signal for downstream disease-prioritization tools.
Cathy Shyr, Yan Hu, Rory J. Tinker, and 9 othersCathy Shyr, Yan Hu, Rory J. Tinker, Thomas A. Cassini, Kevin W. Byram, Rizwan Hamid, Daniel V. Fabbri, Adam Wright, Josh F. Peterson, Undiagnosed Diseases Network, Lisa Bastarache, Hua Xu fewer authors
Vanderbilt; University of Texas; Mount Sinai; Yale
npj Digital Medicine
Thu, Oct 15Multi-omics Single cell biology
Unified variant caller improves structural mutation detection across short- and long-read data
HitSV combines repeat-aware breakpoint detection with haplotype-resolved local DNA assembly to identify structural genomic variants across multiple sequencing platforms. In evaluations on the reference human genome HG002 benchmark, the algorithm improved genotype F1 scores by up to 72% on long-read data and up to 36% on short-read sequences compared to existing tools. Applied to the 1000 Genomes Project phase 4 cohort, the method expanded the catalog of detected structural variants by 31.5%, uncovering widespread composite repeat arrays in human populations.
Gaoyang Li, Yadong Liu, Anqi Zhang, and 5 othersGaoyang Li, Yadong Liu, Anqi Zhang, Xiaoyi Lin, Zheng Hu, Bo Liu, Long Qian, Yadong Wang fewer authors
Nature Biotechnology

Wed, Oct 14Clinical AI Multi-omics
AI scans link brown fat activity to lower psychiatric disease rates and flag candidate drugs
An artificial intelligence system quantified human brown adipose tissue activity across clinical scans, revealing strong associations between active brown fat, improved metabolic profiles, and reduced rates of psychiatric illness.
Nicolás Gómez-Banoy, Jorge A. Barrero, Luiza Rachel Pinheiro de Carvalho, and 15 othersNicolás Gómez-Banoy, Jorge A. Barrero, Luiza Rachel Pinheiro de Carvalho, Caroline S. Jiang, Tara Doma Lama, Nakul Karandikar, Mascha Koenen, Tiffany Lam, Kaja Plucińska, Ksenia Morozova, Stephanie Chahwan, Gustavo Barraza, Burcin Agridag Ucpinar, Jennifer E. Liu, Rohit Sharma, Daniel Lafontaine, Heiko Schöder, Paul Cohen fewer authors
Rockefeller; Memorial Sloan Kettering; Cornell; and 3 othersRockefeller; Memorial Sloan Kettering; Cornell; Laboratorio de Lípidos y Diabetes; Heidelberg; Helmholtz fewer institutions
Nature
Wed, Oct 14Multi-omics Single cell biology
Graph-based method parses tissue architecture across spatial scales
SpatialZoomer uses spectral graph theory to analyze spatial transcriptomics data simultaneously at individual cell resolution and across broad tissue domains.
Xinqi Li, Yuhan Fan, Yue Han, and 10 othersXinqi Li, Yuhan Fan, Yue Han, Wenbo Guo, Jingmin Huang, Nan Yan, Zeyu Chen, Yanhong Wu, Yuxin Miao, Lin Hou, Xuegong Zhang, Zeyu Chen, Jin Gu fewer authors
Tsinghua; China Agricultural University; Peking
Nature Biomedical Engineering
Wed, Oct 14Product launch AI drug discovery
Billion-spectrum proteomics dataset powers fast model for reading peptide sequences
Researchers have released MassNet, an AI-ready training dataset containing 1.5 billion tandem mass spectrometry spectra across 35 species, stored in a tensor format designed for deep learning.
Jun A., Xiang Zhang, Xiaofan Zhang, and 11 othersJun A., Xiang Zhang, Xiaofan Zhang, Jiaqi Wei, Te Zhang, Yamin Deng, Pu Liu, Zongxiang Nie, Yiwen Yu, Yi Chen, Nanqing Dong, Zhiqiang Gao, Siqi Sun, Tiannan Guo fewer authors
Affiliated Hangzhou First People’s Hospital; Westlake Laboratory of Life Sciences and Biomedicine; Westlake University; and 7 othersAffiliated Hangzhou First People’s Hospital; Westlake Laboratory of Life Sciences and Biomedicine; Westlake University; Fudan; University of British Columbia; Shanghai Artificial Intelligence Laboratory; Zhejiang; Westlake Omics Biotechnology Co. Ltd; Soochow University; Shanghai Innovation Institute fewer institutions
Nature Methods
Wed, Oct 14Protein design
Probabilistic algorithm aligns distant protein structures faster and more accurately
LoL-align matches 3D protein structures by iteratively optimizing a local log-odds score based on inter-residue distances.
Lasse Reifenrath, Michel van Kempen, Gyuri Kim, and 6 othersLasse Reifenrath, Michel van Kempen, Gyuri Kim, Soo Hyun Kim, Mohammadreza Radnezhad, Cameron L. M. Gilchrist, Milot Mirdita, Martin Steinegger, Johannes Söding fewer authors
Max Planck; Seoul National University; Korea Basic Science Institute; and 3 othersMax Planck; Seoul National University; Korea Basic Science Institute; Sungkyunkwan University School of Medicine; Biomedical Institute for Convergence at Sungkyunkwan…; Campus Institute Data Science (CIDAS) fewer institutions
Nature Methods
Tue, Oct 13AI drug discovery Multi-omics
Interpretable neural networks spot cancer driver mutations from cellular dependency data
Researchers built two neural network tools, xNNDriver and xAEDriver, to prioritize cancer driver genes by connecting mutations to gene essentiality profiles from the Cancer Dependency Map.
Qingyang Yin, Liang Chen
University of Southern California
PLOS Computational Biology
Tue, Oct 13Multi-omics Single cell biology
Interpretable machine learning decodes how tissues recruit specific cells
A computational framework named SPACER extracts gene-level recruitment rules directly from spatial transcriptomics data to explain how tissues attract specific cell types.
Jia Yao, Ji Seon Shim, Kelly Wong, and 26 othersJia Yao, Ji Seon Shim, Kelly Wong, James Zhu, Yunhe Liu, Sung Wook Kang, Kelli McCord, Bryan M. Burt, Jong Min Choi, Claire Lee, Idania Carolina Lubo Julio, Luisa Maren Solis Soto, Sharia Hernandez, Karen Colbert Maresso, Yang Liu, Yibo Dai, Tian Chu, Wei Lu, Khaja Khan, Paul F. Mansfield, Hee-Jin Jang, William H. Hudson, Tao Yue, Bing Zhang, Changwei Li, Linghua Wang, Hao Zhu, Hyun-Sung Lee, Tao Wang fewer authors
UT Southwestern; Baylor College of Medicine; MD Anderson; University of California, Los Angeles
Nature Biomedical Engineering
Tue, Oct 13Single cell biology Synthetic biology
Computational tool cleans single-cell bacterial genomes to track phages and plasmids
Researchers have developed Scellmate, a computational framework designed to filter contamination and analyze large datasets of single-amplified genomes from single-cell microbial sequencing.
Yanren Wang, Xiaoqing Xu, Xi Chen, and 3 othersYanren Wang, Xiaoqing Xu, Xi Chen, Yulin Zhang, Wenshan Zheng, Tong Zhang fewer authors
University of Hong Kong; Advanced Biomedical Instrumentation Centre
Nature Communications
Mon, Oct 12AI drug discovery Synthetic biology
Interpretable machine learning designs degradable lipids for liver mRNA delivery
Researchers combined interpretable machine learning with modular chemistry to screen degradable ionizable lipids, picking out a lead candidate that successfully delivered mRNA to the liver in mice.
Zhen Zhang, Yongbo Ren, Deqiang Kong, and 14 othersZhen Zhang, Yongbo Ren, Deqiang Kong, Jindong Feng, Kaiyi Hu, Xinyun Xu, Fukang Deng, Peixia Li, Jiuao Wang, Wenhui Huang, Guodan Wei, Yang Su, Ying Tan, Shuixing Zhang, Xiaodan Sun, Feiyu Kang, Hao Yan fewer authors
Tsinghua; Boston University; Georgia Tech; and 3 othersTsinghua; Boston University; Georgia Tech; Harbin Institute of Technology (Shenzhen); Guangxi University; The First Affiliated Hospital of Jinan University fewer institutions
Nature Communications
Sun, Oct 11Foundation model Single cell biology
Vision foundation model adapted to trace cell membranes in noisy cryo-ET volumes
Researchers have adapted a video-segmentation foundation model, named ETSAM, to accurately map cell membranes within noisy 3D cryo-electron tomograms.
Joel Selvaraj, Jianlin Cheng
University of Missouri
Nature Communications
Sun, Oct 11Foundation model Single cell biology
Deep learning extends live-cell super-resolution microscopy duration and speed
Researchers combined two deep-learning approaches to overcome light-induced damage and slow acquisition in RESOLFT super-resolution microscopy.
Guillaume Minet, Anirban Ray, Francesca Pennacchietti, and 3 othersGuillaume Minet, Anirban Ray, Francesca Pennacchietti, Giovanna Coceano, Florian Jug, Ilaria Testa fewer authors
KTH Royal Institute of Technology; Fondazione Human Technopole; Technische Universität Dresden
Nature Communications
Thu, Oct 8
Physics-guided deep learning cleans up noisy fluorescence microscopy from a single frame
Engineers developed an end-to-end deep neural network that restores low-light fluorescence microscopy images and strips out-of-focus background haze using only a single frame.
Xinxiang You, Shuyue Xie, Chengen Li, and 9 othersXinxiang You, Shuyue Xie, Chengen Li, Ke Hu, Xueyang Fu, Cuifang Kuang, Longhua Tang, Ziyi Lu, Chuanhai Fu, Zheng-Jun Zha, Xu Liu, Douguo Zhang fewer authors
University of Science and Technology of China; Zhejiang
PNAS
Thu, Oct 8Single cell biology Virtual cell
Diffusion model predicts unseen genetic perturbation effects and maps causal gene networks
Researchers introduced latent causal diffusion (LCD), a generative diffusion framework designed to model single-cell perturbation responses while accounting for measurement noise.
Lars Lorch, Jiaqi Zhang, Charlotte Bunne, and 3 othersLars Lorch, Jiaqi Zhang, Charlotte Bunne, Andreas Krause, Bernhard Schölkopf, Caroline Uhler fewer authors
ETH Zurich; MIT; Harvard; and 4 othersETH Zurich; MIT; Harvard; Broad Institute; EPFL; Max Planck; European Laboratory for Learning and Intelligent Systems… fewer institutions
PNAS
Thu, Oct 8Product launch
Open-source toolkit extracts readable biological rules from genomic neural networks
Researchers have released Tangermeme, an optimized computational package designed to interpret the complex regulatory logic learned by genomic deep-learning models.
Jacob Schreiber
Research Institute of Molecular Pathology; UMass Chan Medical School
Nature Methods
Thu, Oct 8Protein design
Community benchmark tests 12,000 AI-designed mini-proteins in living CAR T cells
A global competition named Bits to Binders evaluated 12,000 AI-designed 80-amino acid mini-proteins as the targeting heads for CAR T-cell therapies against human CD20.
Clayton W Kosonocky, Alex M Abel, Aaron L Feller, and 129 othersClayton W Kosonocky, Alex M Abel, Aaron L Feller, Amanda E Cifuentes Rieffer, Phillip R Woolley, Jakub Lála, Daryl R Barth, Tynan Gardner, Bits to Binders Competitors Acosta Daniel 6 Angioletti-Uberti Stefano 7 Annapure Vinayak 8 Anthony Neil Barghout Rana A 9 Beining Max 10 Bibekar Parth 11 Biton Dionessa 12 Bryant Patrick 13 Bushuiev Anton 14 Calia Cianna 15 Chan Ava 16 Chen Jie 17 Chenna Akshay 18 Chiang Nicole 19 Chu Vivian 9 Clark Joseph 19 Clark-ElSayed Alia 6 Cole Elliot 20 Cong Qian 17 Das Shreyasi 21 Dean Tanner 19 Derr Tyler 22 Diaz Alejandro 23 Durham Jesse 17 Ertelt Moritz 10 Fonseca Antonio 24 Furman Ora 25 Gardner Tynan 6 Gaza Jokent 26 Greenwood Nathaniel 27 Guo Rui 17 Gurung Abel 28 Haas Vivian 10 Hashemi Seyed Alireza 29 Hoffmann Dieter 10 Hoian Sofiia 30 Howe Corey 31 XiaoYang Hu Calvin 32 Huang Jinling 33 Ibrahim Nader 23 Ishida Satoshi 6 Javed Ansar Ahmad 21 Javvadi Aswini 34 Jerala Roman 35 Joshi Daksh 34 Kiely Even Kim Haelyn 36 Kleiman Diego 19 Knapp Sarah 25 Koh Huan 8 Kouba Petr 14 Kwon Yong Youn 23 Lainšček Duško 35 Lála Jakub 7 Lee Hakyung 36 Li Qiuzhen 13 Lingner Max 10 Liu Yuxuan 37 Liu Yunchao 22 Liu Qianchen 26 Ljubetič Ajasja 35 Ludaic Marko 23 Lundquist Karl Makani Ashish Marchena-Hurtado Javier 38 Martin Freddie 39 Meiler Jens 22 Meireles Fernando 11 Miller David 7 Mitjavila Nuria 23 Mlakar Tjaša 35 Mondal Rajarshi 40 Moradvandi Saeideh 8 Moretti Rocco 22 Mure Sai Jaideep Reddy 37 Naikare Pragati 37 Nasiri Fatemeh 29 Nasrollahi Kaveh 29 Noa Yisel Martinez 26 Openy Joseph 41 Otani Aoi 32 Parmar Vikrant 8 Pei Jimin 17 Perez Alberto 26 Punia Rajat Requena Francisco 42 Rieger Dominic 10 Sahu Aparna 26 Sanchez Elias 23 Satija Rohit 43 Satler Tadej 35 Schlegel Tom 10 Shen Yang 37 Shukla Diwakar 19 De Silva Rangana 26 Singer Benedikt 11 Singh Arjun 44 Singh Amardeep 9 Singh Bhumika 26 Song Jinwoong 36 Srinivasan Harish 17 Stewart Hannah 45 Su Zhaoqian 22 Varga Julia 25 Wang Xin 22 Whitehead Dion 46 Williams Matthew 47 Witwer Max 22 Yin Rujie 37 Yin Song 19 Yuan Rongqing 17 Zadorozhny Nikola 14 Zaveri Dishant Parag 37 Zheng Genhui 6 Zheng Yizhen 8 Zhou Jian 17 Zhu Shaowen 37 Zhu Lin 19, Stephen C Ekker, Andrew D Ellington, Wesley A Wierson, Edward M Marcotte, Daniel Acosta, Stefano Angioletti-Uberti, Vinayak Annapure, Neil Anthony, Rana A Barghout, Max Beining, Parth Bibekar, Dionessa Biton, Patrick Bryant, Anton Bushuiev, Cianna Calia, Ava Chan, Jie Chen, Akshay Chenna, Nicole Chiang, Vivian Chu, Joseph Clark, Alia Clark-ElSayed, Elliot Cole, Qian Cong, Shreyasi Das, Tanner Dean, Tyler Derr, Alejandro Diaz, Jesse Durham, Moritz Ertelt, Antonio Fonseca, Ora Furman, Tynan Gardner, Jokent Gaza, Nathaniel Greenwood, Rui Guo, Abel Gurung, Vivian Haas, Seyed Alireza Hashemi, Dieter Hoffmann, Sofiia Hoian, Corey Howe, Calvin XiaoYang Hu, Jinling Huang, Nader Ibrahim, Satoshi Ishida, Ansar Ahmad Javed, Aswini Javvadi, Roman Jerala, Daksh Joshi, Even Kiely, Haelyn Kim, Diego Kleiman, Sarah Knapp, Huan Koh, Petr Kouba, Yong Youn Kwon, Duško Lainšček, Jakub Lála, Hakyung Lee, Qiuzhen Li, Max Lingner, Yuxuan Liu, Yunchao Liu, Qianchen Liu, Ajasja Ljubetič, Marko Ludaic, Karl Lundquist, Ashish Makani, Javier Marchena-Hurtado, Freddie Martin, Jens Meiler, Fernando Meireles, David Miller, Nuria Mitjavila, Tjaša Mlakar, Rajarshi Mondal, Saeideh Moradvandi, Rocco Moretti, Sai Jaideep Reddy Mure, Pragati Naikare, Fatemeh Nasiri, Kaveh Nasrollahi, Yisel Martinez Noa, Joseph Openy, Aoi Otani, Vikrant Parmar, Jimin Pei, Alberto Perez, Rajat Punia, Francisco Requena, Dominic Rieger, Aparna Sahu, Elias Sanchez, Rohit Satija, Tadej Satler, Tom Schlegel, Yang Shen, Diwakar Shukla, Rangana De Silva, Benedikt Singer, Arjun Singh, Amardeep Singh, Bhumika Singh, Jinwoong Song, Harish Srinivasan, Hannah Stewart, Zhaoqian Su, Julia Varga, Xin Wang, Dion Whitehead, Matthew Williams, Max Witwer, Rujie Yin, Song Yin, Rongqing Yuan, Nikola Zadorozhny, Dishant Parag Zaveri, Genhui Zheng, Yizhen Zheng, Jian Zhou, Shaowen Zhu, Lin Zhu fewer authors
University of Texas; The BioML Society; LifEngine Animal Health (LEAH) Laboratories Incorporated; and 41 othersUniversity of Texas; The BioML Society; LifEngine Animal Health (LEAH) Laboratories Incorporated; Imperial College; Monash; University of Toronto; Leipzig University; EPFL; University of Oldenburg; Stockholm University; Czech Technical University in Prague; University of California, San Diego; Oxford; UT Southwestern; Virtue Therapeutics; University of Illinois; McGill; Sungkyunkwan University; Vanderbilt; Uppsala; Gensaic; Hebrew University; University of Florida; University of Washington; Purdue; University of Tehran; Taras Shevchenko National University of Kyiv; Salt Lake City; Harvard; Aibio AB; Boltzmann Labs; National Institute of Chemistry; Seoul National University; Texas A&M University; Charité; University of Copenhagen; IIIT Delhi; TU Dortmund; Cornell; Pivotal Life Sciences; University of Pittsburgh; University of Michigan; Astera Institute; Northwestern fewer institutions
Molecular Systems Biology
Wed, Oct 7Protein design Synthetic biology
Machine learning guides directed evolution to speed up plant Rubisco
Researchers used the structure-informed protein language model ESM-IF1 to pick mutation sites in tobacco Rubisco, the primary enzyme plants use to fix carbon dioxide.
Julie L. McDonald, Jiacheng Lin, Yunlong Zhao, and 7 othersJulie L. McDonald, Jiacheng Lin, Yunlong Zhao, Brian L. Hie, Rosemary Birch, Mary Gehring, Bryan D. Bryson, Spencer M. Whitney, Matthew D. Shoulders, Robert H. Wilson fewer authors
MIT; Stanford; Australian National University; and 7 othersMIT; Stanford; Australian National University; Whitehead Institute; Howard Hughes Medical Institute; Harvard; Ragon Institute; Massachusetts General Hospital; Broad Institute; Koch Institute fewer institutions
Science Advances
Tue, Oct 6
Computational tool turns light-absorption spectroscopy into quantitative measurements of RNA and DNA shapes
Researchers built NA-CDQuant, a computational framework that deconvolves complex circular dichroism spectra to determine the structural shapes of nucleic acids in solution.
Kevin Mosca, Søren Vrønning Hoffmann, Nykola C. Jones, and 4 othersKevin Mosca, Søren Vrønning Hoffmann, Nykola C. Jones, Richard R. Sinden, Frank Wien, Sergio Marco, Véronique Arluison fewer authors
CNRS; Synchrotron SOLEIL; Sanofi; and 3 othersCNRS; Synchrotron SOLEIL; Sanofi; Aarhus; South Dakota School of Mines and Technology; Université Paris fewer institutions
PNAS
Tue, Oct 6Single cell biology Virtual cell
Optimal transport model infers gene networks and missing protein levels from single-cell snapshots
CardamomOT combines optimal transport mathematics with a mechanistic model of gene expression to infer gene regulatory networks and reconstruct unmeasured protein trajectories across single-cell RNA sequencing time series.
Yann Maugé, Elias Ventre
Aix-Marseille Université
PLOS Computational Biology
Tue, Oct 6
Deep learning identifies disease patterns in spatial biology data without cell labels
Researchers built variational inference-based microniche analysis, or VIMA, a deep learning statistical method that detects disease-linked tissue structures directly from spatial molecular data.
Nature Methods
Mon, Oct 5AI drug discovery
Causal knowledge graphs improve multi-endpoint chemical toxicity predictions
Researchers built a deep-learning framework that embeds directed causal graphs mined from toxicology literature directly into predictive models, mapping chemical structures to multi-target perturbation profiles.
Haoyue Tan, Tong Bao, Yin Fang, and 18 othersHaoyue Tan, Tong Bao, Yin Fang, Rong Zhang, Jinsha Jin, Dan Xu, Lan Xie, Huan Zhong, Xuezhi Xiao, Huixiao Hong, Emilio Benfenati, Tadahaya Mizuno, Qing Zhou, Jingfan Qiu, Changsheng Qu, Yan Mao, Xiangyi Yu, Jing Guo, Hongxia Yu, Xiaowei Zhang, Wei Shi fewer authors
Nanjing University; Jiangsu Province Ecology and Environment Protection Key…; Zhejiang; and 8 othersNanjing University; Jiangsu Province Ecology and Environment Protection Key…; Zhejiang; Foreign Environmental Cooperation Centre; FDA; Istituto di Ricerche Farmacologiche Mario Negri Istituto…; University of Tokyo; The Institute of Statistical Mathematics; Nanjing Medical University; Jiangsu Environmental Engineering Technology Co, Ltd; Solid Waste and Chemicals Management Center fewer institutions
PNAS
Mon, Oct 5AI drug discovery
Machine learning predicts antibiotic synergy while exposing data leakage in earlier models
Researchers developed HALO, a machine-learning framework trained on 3,160 drug-pair-strain interactions across 97 antibacterial compounds and 10 bacterial strains to predict whether pairs of antibiotics work better together.
Hannie Yousefabadi, Mahya Mehrmohamadi
University of Tehran
PLOS Computational Biology
Fri, Oct 2
Algorithm deconvolves dynamic RNA structures from direct sequencing data
A computational method called DeCoRE clusters direct RNA sequencing reads to separate multiple coexisting conformations at single-nucleotide resolution.
Yong Huang, Jinsong Zhang, Lei Sun, Qiangfeng Cliff Zhang
Tsinghua; Academy of Military Medical Sciences; Shandong University
Science Advances
Fri, Oct 2Protein design
Bayesian method accounts for experimental noise in protein fitness training data
Researchers built FLIGHTED, a Bayesian statistical method that converts noisy high-throughput assay readouts into probabilistic fitness landscapes before feeding them into machine-learning models.
Vikram Sundar, Boqiang Tu, Lindsey Guan, Kevin Esvelt
MIT
Cell Systems
Fri, Oct 2Protein design
Generative model links protein sequence and 3D structure to track evolution
Researchers combined amino acid alignments with a structural alphabet, using variational autoencoders and evolutionary energy calculations to map protein sequence variations against structural changes.
Divyanshu Shukla, Jonathan Martin, Faruck Morcos, Davit A. Potoyan
Iowa State University; University of Texas
PNAS
Fri, Oct 2Clinical AI
Brain-metabolism scans allow machine learning to predict tau pathology in dementia
Researchers built a machine learning model that analyzes glucose metabolism in the brain to predict Alzheimer's-type tau pathology across aging and diverse dementia disorders.
Nick Corriveau-Lecavalier, Gemeng Zhang, Hugo Botha, and 27 othersNick Corriveau-Lecavalier, Gemeng Zhang, Hugo Botha, Leland R. Barnard, Heather J. Wiste, Ellen Dicks, Venkatsampath R. Gogineni, Jeyeon Lee, Irene Sintini, Matthew L. Senjem, Christopher G. Schwarz, Petrice M. Cogswell, Derek R. Johnson, Jeffrey L. Gunter, Michael G. Kamykowski, Jonathan Graff-Radford, Vijay K. Ramanan, Julie A. Fields, Mary M. Machulda, Nikki H. Stricker, Bradley F. Boeve, Val J. Lowe, Kejal Kantarci, David S. Knopman, Jennifer L. Whitwell, Keith A. Josephs, Ronald C. Petersen, Clifford R. Jack, David T. Jones, or the Alzheimer’s Disease Neuroimaging Initiative fewer authors
Mayo Clinic; University of Amsterdam; Hanyang University
Nature Communications
Thu, Oct 1Foundation model
Protein language model reconstructs mitochondrial evolution across 200 species
A team retrained a protein language model classifier on experimentally mapped proteomes from eight diverse species to predict the mitochondrial proteins of roughly 200 eukaryotes.
Michael Z Chen, Jonathan A Stefely, Eric G Bardon, and 3 othersMichael Z Chen, Jonathan A Stefely, Eric G Bardon, MitoCarta Tree of Life Consortium, Sarah E Calvo, Vamsi K Mootha fewer authors
Harvard; MIT; Broad Institute; and 3 othersHarvard; MIT; Broad Institute; Howard Hughes Medical Institute; Massachusetts General Hospital; Boston University fewer institutions
Cell
Thu, Oct 1
Two-step machine learning model predicts molecule separation times across different lab setups
Researchers have developed 2-step, a computational framework that predicts small-molecule liquid chromatography retention times without requiring training data from the target system.
Fleming Kretschmer, Eva-Maria Harrieder, Michael Witting, Sebastian Böcker
Friedrich Schiller University Jena; Helmholtz; Labor Becker MVZ eGbR; TU München
Nature Methods
Thu, Oct 1
Standard benchmarks mask real gains from deep learning in genetic perturbation models
Common benchmarking metrics such as mean squared error and Pearson delta fail to detect genuine signal when evaluating models that predict the effects of genetic perturbations, according to an analysis across 14 datasets and 18 evaluation metrics.
Henry E. Miller, Gabriel M. Mejia, Francis J. A. Leblanc, and 3 othersHenry E. Miller, Gabriel M. Mejia, Francis J. A. Leblanc, Brendan Swain, Bo Wang, Lucas Paulo de Lima Camillo fewer authors
Shift Bioscience Ltd; University of Toronto; University Health Network; and 2 othersShift Bioscience Ltd; University of Toronto; University Health Network; Vector Institute; Cambridge fewer institutions
Nature Biotechnology

Publications

Stay ahead of the field.