The importance of this👇cannot be underscored: #caner is a unique evolutionary process. tumors are expected to be genetically v distinct. each tumor = its own world. Therefore: #scRNAseq analyses of cancer tissue data SHOULD NOT‼️BLINDLY include data integration as 1st step🧵

one important reason we haven't reached an "alphafold moment" for biological data beyond protein structures: datasets for protein structure are consistent and uniform. x-ray crystallography or cryo-EM data provide absolute atomic coordinates, meaning that the data is

At @SageBio we believe in the central role of Data Coordinating Centers here. We help build @NCIHTAN and AD Knowledge Portal. There remains tons of scope for increasing incentives on all sides and building performant rather than performative solutions with foundational models in mind.

Guilty as charged. 🧑‍⚖️

100%! Access to annotated data is the key to enterprise AI. The key though is that this AI-ready data is not one-size-fits all- it’s annotated *for each specific use case / operating setting*. This game is not about data lakes… it’s about scalable data labeling & development

One of the disadvantages of coming into a leadership role too soon is that it might stall growth and learning depending on the person. It’s a well known fact that startups inflate titles, and I have seen a few colleagues go from postdoc directly to head of xyz and immediately

🔥 Great news! Attention fellow spatial biologists and pathology-enthusiasts! 🔬 There is a fantastic opportunity for you to join The NIH Innovation Lab "FROM ATLAS TO PRACTICE" September 23rd to 27th where you can help shape how large-scale spatial atlas research can be used to

Mandatory security and privacy training in 2024: Zero content or quiz questions on 2FA or LLMs, but a whole section on etiquette when using a fax machine.

I'm super excited to share Nimbus, our algorithm for automatic classification of cellular marker expression in multiplexed image data! Nimbus is a pretrained deep learning model that works out of the box across panels with any combination of markers (1/x) biorxiv.org/content/10.110…

This is an amazing essay worth reading by anyone applying AI/ML in science. AI/ML has tremendous potential to accelerate biomedical science and clinical research, but it needs to be applied in a thoughtful way just like all our scientific tools.

Friends don’t let friends use non utf-8 encoding in their TIFF tags.

A Tweetorial on "Why do patients with cancer die?" Probably the most important question we need to be able to answer and very nicely addressed in this insightful publication: nature.com/articles/s4156… And which I will now summarise in the following Tweets

The mantra that spatial transcriptomics is about location, location, location is catchy, but what does it really mean? We have just posted biorxiv.org/content/10.110…, work of Kayla Jackson et al., that describes the concordex method for identifying spatial homogeneous regions. 1/🧵

Sharing this cool collaboration with Zwicker Group out today in Science Advances! Using the 3D spatial information of >1,000 precancers from 48 human pancreas samples, we developed a mathematical growth model for pancreatic precancer growth: science.org/doi/10.1126/sc…

Been a little quite on my end of late but there is all sorts of fun things in the works:) The big news🚨 is that CZ #CellxGene is branching out! SUPER excited to share new capabilities for exploring, analyzing, and visualizing spatial transcriptomics data 🚀.

Congratulations Rong Fan! We at the HTAN DCC are excited to work with you!

We at the HTAN DCC are thrilled and excited to work with you!

💡Thinking of using scRNA-seq to design antibody panels for spatial or single-cell profiling? Happy to launch Cytomarker, an interactive web-app and new antibody screening strategy to design and validate custom panels 1/6

A huge day for spatial omics of cancer, tracing the precise origin and evolution of 20 types of cancer, many in 3D, in ~2,000 individuals, with 12 publications nature Nature Cancer Nature Medicine Nature Methods Communications Biology Collection nature.com/immersive/d428…

The first known histological description of precursor lesions of #PancreaticCancer was ~120 years ago (in 1905 to be precise). In this paper, the hand drawn illustrations by Dr. Hulst bear a striking resemblance to lesions that we would today call "pancreatic intraepithelial