Atera whole-transcriptome profiling maps immune responses across tumor clones in DCIS. We reconstruct clonal architecture in situ, validate it with DNA sequencing and track how immunity adapts as tumors evolve. First independent evaluation of the platform
biorxiv.org/content/10.64898…
Prefer to listen? Audio version of the paper, via Paper2Pod:
paper2pod.com/share?pid=2423…
Mapping fine-grained immune annotations onto reconstructed clonal architecture shows how the microenvironment changes as individual clones expand - a framework for tumor-immune coevolution in archival patient tissue.
Very proud of this work, led by @Matusiak
We laser-microdissected 16 tumor regions and sequenced them to confirm copy-number alterations inferred from transcriptomics. First time single-cell spatial WTA copy-number calls have been validated against direct DNA measurement. The inferred clones are real.
Atera captures what targeted panels miss: rare cell states and functional populations invisible to fixed gene panels. 18,000 genes vs 280-5,000 on Xenium - whole-transcriptome at single-cell resolution, tumor architecture preserved, genome-scale inference enabled. @10xGenomics
david van ijzendoorn retweeted
Wanna learn abt spatial #macrophage niches in human? Listen to our paper in a form of #AI generated podcast #Paper2pod by @vanijzen. Link below
david van ijzendoorn retweeted
Tired of reading your papers? Transform them into a podcast with #Paper2Pod. Great idea @vanijzen
david van ijzendoorn retweeted
It seems I also have to try Paper2Pod
Paper2Pod is a free iPhone app that turns research papers into audio podcasts 👇
NotebookLM app also coming!😊
david van ijzendoorn retweeted
NotebookLM generates AI podcasts but you can't use it on the phone.
Paper2Pod is a free iPhone app that turns research papers into audio podcasts, so you can listen to them on the go.
Here's how to use it:
david van ijzendoorn retweeted
The Spatial Human #Macrophage Map is finally online in @CD_AACR! We show how phenotypically distinct macrophage populations occupy spatially segregated tissue #niches, with #IL4I1 and #SPP1 TAMs predicting opposite outcomes in colon cancer!
david van ijzendoorn retweeted
✅Instead of ChatGPT, use apps like Scite(dot)ai, Consensus(dot)app, EvidenceHunt(dot)com, and The-Literature(dot)com.
These apps will answer your questions with references to published papers.
See how Scite answers my question and cites real articles👇
david van ijzendoorn retweeted
Evidence Hunt
Evidencehunt.com via @vanijzen et al
I’ve just come across an online AI-powered FREE chatbox answers clinical Qs EXCLUSIVELY based on NICE guidelines or PubMed.
Although the idea might sound simple for tech savvies but as a clinician, I liked this tailored approach in developing AI tools for healthcare #MedTwitter
It's cool to see that people are finding our chat with guidelines useful!
Replying to @MustafaAjlan
Not sure about the validation work made around this but tried it on several topics and provided me with good answers. Evidencehunt.com
david van ijzendoorn retweeted
You can also look up articles based on Population, Intervention, and Study Type.
Click on "Search" and fill in the fields of Population and Intervention and select a Study Type.
david van ijzendoorn retweeted
Evidence Hunt is an AI-powered that will answer all your biomedical questions with references to published articles.
It's like having a ChatGPT for PubMed and NICE* guidelines. And it's totally FREE!
*National Institute for Health and Care Excellence, UK
Go to evidencehunt(dot)com and log in to your account. Sign up if you don't have one.
Simply type in your question and Evidence Hunt will answer it with references to published articles.
Click on a cited source and Evidence Hunt will give you its PICO (Population, Intervention, Control, and Outcomes) analysis.
If you want answers based on NICE guidelines, simply click on "NICE Guidelines" and ask your question.
In the example below, I asked Evidence Hunt if I could use a belt to self manage lower back pain.
It said that according to NICE guidelines, belts should not be used to self manage lower back pain.
It's great to see Dr. Bilal feature EvidenceHunt! Definitely follow @MushtaqBilalPhD if you're interested in learning how to boost your research productivity.
Evidence Hunt is an AI-powered that will answer all your biomedical questions with references to published articles.
It's like having a ChatGPT for PubMed and NICE* guidelines. And it's totally FREE!
*National Institute for Health and Care Excellence, UK
Go to evidencehunt(dot)com and log in to your account. Sign up if you don't have one.
Simply type in your question and Evidence Hunt will answer it with references to published articles.
Click on a cited source and Evidence Hunt will give you its PICO (Population, Intervention, Control, and Outcomes) analysis.
If you want answers based on NICE guidelines, simply click on "NICE Guidelines" and ask your question.
In the example below, I asked Evidence Hunt if I could use a belt to self manage lower back pain.
It said that according to NICE guidelines, belts should not be used to self manage lower back pain.
david van ijzendoorn retweeted
New technology has always generated fear amongst people, even famous scholars.
In fact, Conrad Gessner (1516-1565), a famous Swiss physician was against the PRINTING PRESS.
At the time, there were 10,000 book titles published by the printing press.
This shocked and worried Gessner as he believed that ordinary people could not handle so much knowledge.
He demanded European countries to enforce a law that regulates the sale and distribution of books to ordinary people.
New technology has and always will create fear. @lexfridman
Examining the cost differences for 1,000 tokens/5,000 chars between Palm2 and 3.5 Turbo:
Input:
- Palm2 = 0.0005
- OpenAI = 0.0075
- 15x price difference
Output:
- Palm2 = 0.0005
- OpenAI = 0.01
- 20x price difference
But the output disappoints..
Closing my chapter as a Stanford postdoc. Excited to dive full-time into my startup journey! 🚀 #StartupLife #StanfordToStartup
#McKinsey study shows a ~40% decrease in code generation time using generative AI. I see the same efficiency improvement in my #Bioinformatics and #DataScience workflows. For those using R, here's a quick setup:
1. Use Cursor as your IDE.
2. Install the Jupyter extension.
3. Install IRkernel in R.
4. Create a notebook (.ipynb) file and select your R kernel.
5. Press cmd-k to generate code
mckinsey.com/capabilities/mc…













