[{"data":1,"prerenderedAt":152},["ShallowReactive",2],{"content-query-Ix5FDPSzAj":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":146,"_id":147,"_source":148,"_file":149,"_stem":150,"_extension":151},"/news/zh/3855","zh",false,"","Nature子刊收录，部分性能超越AlphaFold3！基于昇思打造的蛋白质复合物结构预测模型GRASP亮相","这项研究不仅突破了AlphaFold系列的部分技术瓶颈，也丰富了实验科学与AI协同的新范式，未来将有望在生物医药领域取得重要应用。","2025-10-11","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/10/13/4506cd7a160d475a92f48e0c5b69bc53.png","news",{"type":14,"children":15,"toc":143},"root",[16,24,30,35,44,49,54,62,78,83,88,93,98,105,119,124,129,134,139],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"nature子刊收录部分性能超越alphafold3基于昇思打造的蛋白质复合物结构预测模型grasp亮相",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":23,"value":29},"2025年9月，国际顶级期刊《Nature Methods》发表了来自北京昌平实验室、北京大学前沿交叉学科研究院等团队的重磅研究成果——蛋白质复合物结构预测模型GRASP。这项研究由高毅勤教授团队领衔，联合北京大学生命科学学院、化学与分子工程学院及华为技术有限公司昇思MindSpore AI4S团队共同完成，论文的发表标志着中国科学家在AI+生物计算领域取得突破性新进展。",{"type":17,"tag":25,"props":31,"children":32},{},[33],{"type":23,"value":34},"论文地址：",{"type":17,"tag":25,"props":36,"children":37},{},[38],{"type":17,"tag":39,"props":40,"children":42},"a",{"href":41},"nature.com/articles/s41592-025-02820-1",[43],{"type":23,"value":41},{"type":17,"tag":25,"props":45,"children":46},{},[47],{"type":23,"value":48},"该论文提出的GRASP模型通过直接整合多种实验约束信息，显著提高了蛋白质复合物结构预测的准确性和可靠性。GRASP在处理模拟和真实世界的实验约束方面表现出色，特别是在抗体-抗原复合物预测中超越了现有的AlphaFold3。",{"type":17,"tag":25,"props":50,"children":51},{},[52],{"type":23,"value":53},"GRASP还可以作为综合建模工具，结合多种实验信息进行复杂的约束建模。未来的工作将进一步优化GRASP的集成策略，以解决更具挑战性的应用场景。",{"type":17,"tag":25,"props":55,"children":56},{},[57],{"type":17,"tag":58,"props":59,"children":61},"img",{"alt":7,"src":60},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/10/13/6a817c12eeaa4c708e7b08579ae80941.png",[],{"type":17,"tag":25,"props":63,"children":64},{},[65,71,73],{"type":17,"tag":66,"props":67,"children":68},"strong",{},[69],{"type":23,"value":70},"# 01",{"type":23,"value":72}," ",{"type":17,"tag":66,"props":74,"children":75},{},[76],{"type":23,"value":77},"核心创新：实验数据与AI的深度耦合",{"type":17,"tag":25,"props":79,"children":80},{},[81],{"type":23,"value":82},"GRASP的核心突破在于首创\"实验约束引导AI预测\"的范式：",{"type":17,"tag":25,"props":84,"children":85},{},[86],{"type":23,"value":87},"1. 数据转化技术：将交联质谱（XL-MS）、深度突变扫描（DMS）等稀疏实验数据转化为残基对限制（RPR）和界面限制（IR），通过修改AlphaFold-Multimer的注意力机制实现精准嵌入 。",{"type":17,"tag":25,"props":89,"children":90},{},[91],{"type":23,"value":92},"2. 抗噪系统：开发迭代噪声过滤算法，可自动识别并剔除错误实验数据（如误差超过30%的XL-MS数据） 。",{"type":17,"tag":25,"props":94,"children":95},{},[96],{"type":23,"value":97},"3. 多模态融合架构：支持冷冻电镜密度图、化学位移扰动等6类实验数据混合输入，首次实现\"低分辨率实验+高精度AI\"的协同建模 。",{"type":17,"tag":25,"props":99,"children":100},{},[101],{"type":17,"tag":58,"props":102,"children":104},{"alt":7,"src":103},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/10/13/de43bbb014574ce18b04ce3051bd2f54.png",[],{"type":17,"tag":25,"props":106,"children":107},{},[108,113,114],{"type":17,"tag":66,"props":109,"children":110},{},[111],{"type":23,"value":112},"# 02",{"type":23,"value":72},{"type":17,"tag":66,"props":115,"children":116},{},[117],{"type":23,"value":118},"部分创新场景性能倍级领先AlphaFold3",{"type":17,"tag":25,"props":120,"children":121},{},[122],{"type":23,"value":123},"GRASP 在部分公开测试中表现优异：",{"type":17,"tag":25,"props":125,"children":126},{},[127],{"type":23,"value":128},"1. 在界面基准测试中，GRASP方法表现全面最佳。其效率极高，仅需引入两个跨链接触约束，就能将DockQ评分提升至0.35，并使过半样本达到可接受水平。",{"type":17,"tag":25,"props":130,"children":131},{},[132],{"type":23,"value":133},"2. GRASP在XL、CL、CSP约束数据实验中，对低覆盖率或高噪声的各类实验数据中均表现稳健，性能显著优于其他方法。",{"type":17,"tag":25,"props":135,"children":136},{},[137],{"type":23,"value":138},"3. 在抗原-抗体预测上，GRASP在深度突变扫描数据中展现出卓越性能与强鲁棒性，预测成功率与精度均大幅领先AlphaFold3结果23%-38%。",{"type":17,"tag":25,"props":140,"children":141},{},[142],{"type":23,"value":9},{"title":7,"searchDepth":144,"depth":144,"links":145},4,[],"markdown","content:news:zh:3855.md","content","news/zh/3855.md","news/zh/3855","md",1776506091030]