[{"data":1,"prerenderedAt":105},["ShallowReactive",2],{"content-query-LASmtq3OUD":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":99,"_id":100,"_source":101,"_file":102,"_stem":103,"_extension":104},"/news/en/1442","en",false,"","New Protein Structure Prediction Model Running on MindSpore Ranks First in CAMEO","The MindSpore-based model in protein structure prediction took first place in CAMEO for three consecutive weeks worldwide.","2022-04-24","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/04/26/909e2efa2e944c3b81a43c0be0cc9daf.png","news",{"type":14,"children":15,"toc":96},"root",[16,24,30,38,43,48,53,58,63,76,81,86,91],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"new-protein-structure-prediction-model-running-on-mindspore-ranks-first-in-cameo",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":23,"value":29},"[Shenzhen, China, April 21, 2022] Huawei topped the Continuous Automated Model EvaluatiOn (CAMEO) chart for three consecutive weeks thanks to the cutting-edge performance of its MindSpore-based protein structure prediction model. This news demonstrates the powerful capabilities created by Ascend AI software and hardware in the AI + scientific computing scenarios.",{"type":17,"tag":25,"props":31,"children":32},{},[33],{"type":17,"tag":34,"props":35,"children":37},"img",{"alt":7,"src":36},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/04/26/97194f3e93644b68bd1571ad9908e503.png",[],{"type":17,"tag":25,"props":39,"children":40},{},[41],{"type":23,"value":42},"Latest monthly results released on 21st, April",{"type":17,"tag":25,"props":44,"children":45},{},[46],{"type":23,"value":47},"Ranking No. 1 for three consecutive weeks",{"type":17,"tag":25,"props":49,"children":50},{},[51],{"type":23,"value":52},"CAMEO, an event jointly hosted by the Swiss Institute of Bioinformatics and the University of Basel in Switzerland, is one of the most important competitions in the protein structure prediction field. During the event, each team must predict the latest 20 protein structures cracked by structural biologists around the world. The scores and rankings of each team are updated online every week. The competition attracts many outstanding model developers in the biocomputing field worldwide.",{"type":17,"tag":25,"props":54,"children":55},{},[56],{"type":23,"value":57},"The Huawei entry was a collaborative effort with the China Changping Laboratory, Peking University's Biomedical Pioneering Innovation Centre (BIOPIC) and Chemistry and Molecular Engineering College, Shenzhen Bay Laboratory's Gao Yiqin team, and Peng Cheng Laboratory's Chen Jie team. Together, the team successfully developed a model to predict protein structures. The model was trained on the MindSpore AI framework, and took first place in the competition for three consecutive weeks worldwide.",{"type":17,"tag":25,"props":59,"children":60},{},[61],{"type":23,"value":62},"Advances of the MindSpore model for protein structure prediction",{"type":17,"tag":25,"props":64,"children":65},{},[66,68,74],{"type":23,"value":67},"Protein structure prediction is one of the key studies of artificial intelligence (AI) in life sciences, which also makes it of great significance to biological research. In December 2020, the publication ",{"type":17,"tag":69,"props":70,"children":71},"em",{},[72],{"type":23,"value":73},"Science",{"type":23,"value":75}," selected ten major scientific breakthroughs of the year, including AlphaFold, an AI technique used for protein structure prediction that has become a benchmark for biological research.",{"type":17,"tag":25,"props":77,"children":78},{},[79],{"type":23,"value":80},"In November 2021, Huawei jointly launched a tool for protein structure prediction and inference that runs on the algorithms of AlphaFold2. In February 2022, the team streamlined its training process, improving the efficiency by two to three times.",{"type":17,"tag":25,"props":82,"children":83},{},[84],{"type":23,"value":85},"Implemented on the Ascend AI software and hardware platform, it enables mixed precision, slashing single-step iteration from 20 seconds to 12 seconds, increasing performance over 60%. In addition, based on the memory overcommitment capability of MindSpore, its training sequence length is increased from 384 to 512.",{"type":17,"tag":25,"props":87,"children":88},{},[89],{"type":23,"value":90},"The record-breaking prediction result and rank in CAMEO signals a milestone for MindSpore in protein structure prediction, filling a scientific gap in this field domestically. The training accuracy of the MindSpore-based tool is very close to that of AlphaFold2, but that doesn't stop the MindSpore team from improving its algorithms and training scale and optimizing the software and hardware compatibilities.",{"type":17,"tag":25,"props":92,"children":93},{},[94],{"type":23,"value":95},"Looking forward, MindSpore hopes to cooperate with more academic and industry partners to continuously make breakthroughs in protein structure prediction and fuel pharmaceutical development.",{"title":7,"searchDepth":97,"depth":97,"links":98},4,[],"markdown","content:news:en:1442.md","content","news/en/1442.md","news/en/1442","md",1776506042316]