Nature biotechnology · 2026

A blinded, prospective benchmark of in silico antibody discovery anchored to experimental affinity and developability.

СтатьяНазвание и аннотация: Перевод в очереди

Обработка статьи · этапы, ошибки и управление

Загрузка состояния…

История · было → стало

Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.

Erasmus M Frank, Bedinger Daniel, Hopkins Elizabeth, Ferguson Ginger, Strickler Justine, Graff Christilyn P, Summers Samantha R, Capehart Stacy L, Slocum Joshua D, Richardson Crystal, Kumar Sumit, Sun Zhifei, Shang Yujie, Zhang Jixian, Gu Ming, Yi Lixia, Wellner Alon, Zheng Shuangjia, Lu Wei, Sormanni Pietro, Greenig Matthew, Zhang Haiping, Mann Brendan T, Baghbanzadeh Mahdi, Rahnavard Ali, Moore Gregory L, Sun Huaiyu, Ding Ying, Nisthal Alex, Kanodia Jitendra, Bernett Matthew J, Pélissier Aurélien, Shao Yanjun, Rodriguez Martinez Maria, Ramesh Karthik, Nastri Horacio, Evers Andreas, Abdelatif Anhar, Bordner Andrew J, Bordyuh Mykola, Heo Lim, Kidd Brian A, Serhat Tetikol H, Wei Shuai, Shin Jung-Eun, Peckner Ryan, Manley Leigh, Lunge Ajitesh, Devasurmutt Yashas, Guloglu Bora, Copoiu Liviu, McGibbon Miles, Fernandez-Quintero Monica L, Mishra Nitesh, Callaghan Sean M, Swanson Olivia M, Bader Daniel L V, Ferguson James A, Raghavan Sai S R, Nemoz Benjamin, Maillie Colleen A, Bowman Charles, Briney Bryan, Ward Andrew B, Marcatili Paolo, Akbar Rahmad, He Bing, Wu Fandi, Yao Jianhua, Hu Bin, Kucer Michal, Gibson Kaetlyn Rose, Somasundaram Rahul, Hung Li-Wei, Kaszuba Tomasz, Fremont Daved H, Jeong Hyeongsun, Kurella Vinodh Babu, Malhotra Shipra, Kumar Satyendra, Liu Yanyun, Xu Lingling, Misa Joshua, John Alexander Nicholas St, Vogt Jeff, Dávila-Hernández Fátima A, Xu Da, Chungyoun Michael, Huggan Zyaja D, Gray Jeffrey J, Parkinson Jonathan, Ko Young Su, Wang Wei, Geiger Franziska, Ziegler Jonathon D, Haas Nikhil, Challacombe Chance, Qamar Ahmad, Singh Akshita, Tang Yi-Ching, An Zhiqiang, Jiang Xiaoqian, Kim Yejin, Zhao Xinyan, Swanson Erik, Klattig Jürgen, Winkler Karsten, Bataa Tschimegma, Sandig Volker, Denzler Lilian, Liu Chunan, Brezski Randall J, Spector Laura, Perea-Schmittle Katheryn, D'Angelo Sara, Ferrara Fortunato, Bradbury Andrew R M

Аннотация

Experimentally validated prospective, blinded benchmarks are needed to separate durable advances from hype in computational antibody design. Here AIntibody, a challenge inspired by the Critical Assessment of Structure Prediction, tests 511 artificial intelligence (AI)-designed or predicted antibodies from 29 organizations on three tasks: in silico affinity maturation from phase 1 sequencing outputs, affinity ranking within heavy-chain complementarity-determining region 3 (HCDR3) clusters of a selection output and CDR design of proteins not included in a selection output. Validated with diverse experimental assays, several groups produced developable antibodies with affinities <100 pM. However, these successes were exceptions that did not transfer across tasks. Affinity-matured antibodies were modeled effectively. Except for one model, predicting high-affinity clones from clustered HCDR3 datasets was worse than random clone picking. Out-of-library design was highly variable for most method submissions, with many failing to outperform standard selections. The AIntibody challenge shows that AI can optimize antibodies in defined, biologically grounded regimes, in addition to highlighting critical gaps including affinity prediction and library-inspired antibody design and cross-task generalization.

A blinded, prospective benchmark of in silico antibody discovery anchored to experimental affinity and developability.

Ошибки и ограничения

Аннотация