Yuzhou Chang

Chuffed to share KRONOS: A foundation model for Spatial Proteomics! arxiv.org/abs/2506.03373 Exciting work with the indomitable @AI4Pathology, led by the amazing @mshaban_ai, @Yuzhou_Chang, with Huaying, @yeoyaoyu, @GuillaumeJaume, Andrew Song, and our collaborators

@Guangyu_Wang01 congrats! Important tool!

We are so happy to publish our new foundation model in Nature Methods. we trained OmiCLIP, a vision-omics due modalities model to bridge pathology image and transcriptomic, and Loki, a platform using OmiCLIP as backbone for ST and HE image cross analysis. nature.com/articles/s4159…

Multi-agent architectures are the FUTURE Here are 6 different types: 𝟭. 𝗛𝗶𝗲𝗿𝗮𝗿𝗰𝗵𝗶𝗰𝗮𝗹 (𝗩𝗲𝗿𝘁𝗶𝗰𝗮𝗹) 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 A supervisor agent orchestrates multiple specialized agents. 𝘌𝘹𝘢𝘮𝘱𝘭𝘦: • One agent retrieves information from internal data sources • Another agent specializes in public information from web searches • A third agent specializes in retrieving information from personal accounts (email, chat) 𝟮. 𝗛𝘂𝗺𝗮𝗻-𝗶𝗻-𝘁𝗵𝗲-𝗟𝗼𝗼𝗽 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 Incorporates human verification before proceeding to next actions, used when handling sensitive information. 𝟯/𝟱. 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 (𝗛𝗼𝗿𝗶𝘇𝗼𝗻𝘁𝗮𝗹) 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 Agents communicate directly with one another in a many-to-many fashion. Forms a decentralized network without strict hierarchical structure. 𝟰. 𝗦𝗲𝗾𝘂𝗲𝗻𝘁𝗶𝗮𝗹 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 Agents process tasks in sequence, where one agent's output becomes input for the next. 𝘌𝘹𝘢𝘮𝘱𝘭𝘦: Three sequential agents where: • First query agent retrieves information from vector search • Second query agent retrieves additional information from web search based on first agent's findings • Final generation agent creates a response using information from both query agents 𝟱. 𝗗𝗮𝘁𝗮 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 Includes agents dedicated to transforming data. 𝘌𝘹𝘢𝘮𝘱𝘭𝘦: • A transformation agent that enriches data at insert-time or transforms existing collections There are also some other patterns that can be combined with these architectures: • 𝗟𝗼𝗼𝗽 𝗽𝗮𝘁𝘁𝗲𝗿𝗻: Iterative cycles for continuous improvement • 𝗣𝗮𝗿𝗮𝗹𝗹𝗲𝗹 𝗽𝗮𝘁𝘁𝗲𝗿𝗻: Multiple agents working simultaneously on different parts of a task • 𝗥𝗼𝘂𝘁𝗲𝗿 𝗽𝗮𝘁𝘁𝗲𝗿𝗻: A central router determining which agents to invoke • 𝗔𝗴𝗴𝗿𝗲𝗴𝗮𝘁𝗼𝗿/𝘀𝘆𝗻𝘁𝗵𝗲𝘀𝗶𝘇𝗲𝗿 𝗽𝗮𝘁𝘁𝗲𝗿𝗻: Collecting and synthesizing outputs from multiple agents Check out this ebook for more info: weaviate.io/ebooks/agentic…

Multi-node NVIDIA HGX B200-accelerated clusters are available NOW, on-demand through Lambda 1-Click Clusters.

Spatial proteomics is here, but spatial functional proteomics? Here is our Nature BME paper on spatial protein interactomics (nature.com/articles/s4155…) illuminating how 47 proteins co-localize/ interact within 20 nm and their function in tissues. @naturemethods @rita_strack

Model Context Protocol (MCP), clearly explained:

RIP Data Scientists. The Generative AI Data Scientist is NOW what companies want. This is actually good news. Let me explain:

Lastly, I am very excited to integrate GSP, a novel mathematical theory based on graphs with novel types of spatial omics data. Moving forward, we will continue to explore and develop new theoretical models in conjunction with spatial biotechnology. ⏭️ To be continued ⏭️

Fifth, we apply SGCC across multiple samples, obtaining a series of SGCC scores as spatial factors. We then use differential expression analysis or time-series modeling to calculate gene expressions that correlate with these spatial factors. Yao has previously described.

Following Yao's post @yeoyaoyu , I'm very excited to share the principles of spectral graph cross-correlation (SGCC). I would also like to express my gratitude for the support and guidance from my co-mentors @SizunJ and @QinMaBMBL.

@yeoyaoyu @SizunJ @QinMaBMBL teamwork🙌

Spatial proteomics & transcriptomics are amazing technologies that have transformed how we study diseases. Can we do more? Here's our new preprint that combines both assays into one! (massive thanks to @SizunJ @Yuzhou_Chang Huaying @QinMaBMBL et al. biorxiv.org/content/10.110…

Best Practices for Using AI When Writing Scientific Manuscripts (1/3)