近日，随着 “杭州六小龙” 刷屏，新华社、英国《经济学人》等海内外媒体都关注到坐落在杭州的浙江大学。“六小龙” 背后，DeepSeek 创始人梁文锋在浙大信息与电子工程学院完成本硕学习，云深处科技创始人朱秋国毕业、任教于浙大，群核科技三名创始人中的两位——黄晓煌、陈航毕业于浙大竺可桢学院。

与此同时，基础科研领域也传来好消息。具有全球影响力的自然指数（Nature Index）官网更新最新排名，哈佛大学位居全球高校第一，浙江大学位居全球高校第三。（自然指数是依托于《自然》系列、《科学》《细胞》、医学等 145 种全球顶级期刊，统计全球各高校、科研院所（国家）发表论文数量的数据库，是国际公认的衡量自然科学领域高质量研究产出与合作情况的重要指标。）同期，在最新一轮全国重点实验室重组后，浙江大学牵头的实验室数量排名全国第二，仅次于清华大学。

这些在产业界和基础研究领域同步取得的成果背后，有哪些驱动因素？产学研融合发展之路在何方？创新创业如何借力高校科研资源？就上述话题，浙江大学在上周首办上市公司企业家校友交流会，校党委书记任少波出席活动并发表讲话，晶泰科技联合创始人、首席执行官马健博士受邀参与活动，与 30 余名上市公司企业家校友一起回到母校，共同探讨、热议交流。

浙江大学党委书记任少波发表讲话

任少波指出，2025 年会带来新的机遇和挑战。一批师生、校友创新创业的事迹和贡献，让全球浙大人自豪、振奋，要从中学习和弘扬求是之风，坚守本质和长期主义，形塑诚勤实真的浙大人特质；高举 “创新浙大” 旗帜，勇于谋划原创和战略任务，组织强大链接携手走向全球前列；擦亮 “创业浙大” 品牌，提供高品质前瞻性交流平台，让校友会成为战略服务者，在新形势下创新校友工作模式，共筑校友生态网络。

晶泰科技联合创始人、首席执行官马健博士现场发言

马健博士于浙江大学完成本科学习并保送直博，理论物理博士毕业后前往麻省理工学院继续博士后研究。2015 年，马健博士与温书豪博士、赖力鹏博士联合创立晶泰科技，公司自主研发了基于量子物理并以人工智能赋能和机器人驱动的创新研发平台，加速新药研发、新材料发现以及化学自动化，并于 2024 年 6 月在香港交易所主板挂牌上市，成为首家以香港 18C 规则上市的特专科技企业。

马健博士表示，一直以来，晶泰科技都致力于推进基础科研和前沿技术的融合创新，面对新一轮科技革命和产业变革带来的机遇和挑战，将依托 AI + 机器人科技平台型企业的优势，重点在生物医药、新材料、化学化工等垂直领域探索开发超级人工智能。在此过程中，企业将不断加强与浙江大学等基础科研实力雄厚的高校院所在联合技术攻关、科研成果转化、创新人才培养等方面的交流与合作，构建产学研联合创新共同体，以协同创新模式，挑战破解科研难题，共同投身到中国式现代化建设中，参与新一轮全球科技和产业竞争。

浙江大学上市公司企业家校友交流会全体合影

The post 马健博士受邀参加浙江大学上市公司企业家校友交流会 appeared first on 晶泰科技 XtalPi.

挑战:

• SARM1具显著的动态构象变化，体外难以稳定并捕捉其激活态构象；
• 目前的候选药物分子主要为竞争性抑制剂，缺乏明确的抑制机制。

目标:

• 通过晶泰科技AI驱动的Cryo-EM图像处理模块，帮助加速解析抑制剂与SARM1的高分辨率电镜结构，为理性药物设计提供新思路、新方向；
• 进一步揭示该靶点的自抑制态与激活态的转换机制，加深对靶点作用机理的理解。

结构解析流程：

解析结果：

项目小结：

仅使用200kV电镜，在8 周内成功解析SARM1与抑制剂的2.8 Å结构，这项开创性研究成果揭示了一种小分子抑制剂的竞争性抑制机制。高比例NMN/NAD+诱导激活SARM1，抑制剂在SARM1催化下与NAD+发生碱基交换反应，生成抑制剂-ADPR这一新产物，该产物与SARM1结合并锁定其处于激活态构象，封锁底物NAD+结合口袋，从而抑制SARM1的NAD+降解活性。该结果为证实SARM1的激活机制和碱基交换反应提供了最直接的结构证据。

数据计算过程中，通过使用 AI 驱动的图像处理模块实现了目标颗粒的精确挑选，与经典方法相比，在同等算力下，加速整体计算流程并实现更高的分辨率。

同时，首次从分子水平揭示了 SARM1 的激活机制，加深对其如何介导退行性疾病发生机理的理解。

The post 8周解析SARM1蛋白结构 appeared first on 晶泰科技 XtalPi.

Explore Two Case Studies on How XtalPi Leveraged MicroED to Address Different Polymorphism Challenges

Structural elucidation is a critical process in understanding and addressing polymorphism challenges in drug development. Here are two case studies highlighting how our technical team employed MicroED to determine the structures of crystalline samples in just 5 days, providing crucial structural information to tackle complex polymorphism issues.

Case Study 1: Determining Component Ratio in a Non-stoichiometric Co-former System Using MicroED 

XtalPi’s Approach & Findings

• Explained the varying stoichiometric ratios of a monobasic API and methanesulfonic acid in a non-stoichiometric co-former through MicroED-derived crystal structure.
• Revealed that methanesulfonic acids are situated in disordered lattice channels within a complex asymmetric unit that contains 8 APIs and 14 co-formers.

Case Study 2: MicroED Discovered an Unexpected Polymorphism from Samples with Highly Similar XRPD Patterns

XtalPi’s Approach & Findings

• Identified distinct polymorphic structures from two batches of samples with nearly identical XRPD spectra but different analytical profiles.
• Mitigated polymorphism risk by achieving definitive structural assignments using MicroED when single crystals could not be obtained for the samples. 

The post How XtalPi Utilized MicroED to Tackle Different Polymorphism Challenges appeared first on 晶泰科技 XtalPi.

Case Study: Harnessing CSP and MicroED to Determine Remdesivir’s Most Stable Polymorph in Just 33 Days

Accurately identifying the most stable polymorph is essential for successful drug development, especially when facing tight R&D timelines and limited material availability. Faced with two similar Remdesivir anhydrates, our team utilized an integrated experimental and AI-driven computational approach to quickly pinpoint the most stable polymorph across varying temperatures, ensuring a reliable drug product development. 

Here’s how we delivered results 2x faster than standard polymorph research:

• Utilized MicroED to determine crystal structures of Form II and, for the first time, of Form IV to explain their similar XRPD patterns and properties. 
• Leveraged AI-powered Crystal Structure Prediction (CSP) platform to predict and recommend Form II as the most stable anhydrate at room and high temperatures, well ahead of traditional timelines. 

The post Accelerated Identification of the Most Stable Remdesivir Polymorph in 33 Days Using an Integrated Experimental and Computational Approach appeared first on 晶泰科技 XtalPi.

Case Study: How the XtalPi Team Discovered Novel, Non-Covalent, Potent Hits Against GPX4 in 28 Days

GPX4 is a critical enzyme for cellular antioxidant defense, but is a highly challenging target. In this case study, we share the story of how we discovered three novel non-covalent hit compounds with biochemical IC₅₀ < 10µM in just 28 days.

• Using predictive AI models and physics-based computations, we probed key pharmacophores with high accuracy and throughput, even without reference compounds.
• 159 library compounds were proposed and 124 were successfully synthesized using our automation platform.

The post Hit Identification – Novel Hits for GPX4 appeared first on 晶泰科技 XtalPi.

The post Application Note: DEL Screening appeared first on 晶泰科技 XtalPi.

Case Study: How the XtalPi Team Discovered a Selective, Novel Lead Series for Key Cancer Target SMARCA2

Loss-of-function mutations in SMARCA4 occur in various cancers, leading to an increased dependency on the structurally similar SMARCA2 for activity. This makes SMARCA2 an attractive therapeutic target due to its increased importance in cancer cell survival.

However, identifying novel small-molecule SMARCA2-selective inhibitors poses significant challenges. To address this, XtalPi leveraged its advanced AI-powered discovery platform and Cryo-EM technology, opening up new possibilities for targeted cancer treatments.

Our Key Results: 

• Discovered a novel lead series with high efficacy and 20-fold selectivity for SMARCA2 over SMARCA4 in biochemical assays, showcasing strong potential for further development.
• Identified a unique SMARCA2 binding pocket using Cryo-EM at 2.9 Å resolution, offering new insights for the design of selective inhibitors and advancing structure-based drug discovery.
• Achieved 100% library compound synthetic success rate by proprietary synthetic feasibility prediction models.

The post Lead Identification – Discovering Selective Inhibitors Against an Unreported Binding Pocket of SMARCA2 via a Rational Drug Design Approach appeared first on 晶泰科技 XtalPi.