Shantanu Singh

Ep 23 - So what actually IS CellPainting anyway? You may have heard of it if you're say, a fan of the @drannecarpenter.bsky.social - @shantanu-singh.cc lab (and who isn't????), but what, you know, is it? Let @erinweisbart.bsky.social break it down for you!

youtu.be/9-d6DoHtgUY

💡Just presented our new paper at the @iccv.bsky.social BioImage Computing workshop: cubic: CUDA-accelerated 3D Bioimage Computing. We introduce a simple way to add GPU acceleration to scikit-image–based bioimage processing pipelines by swapping import statements. 🧵
1/6 #iccv2025

Morph Map is now published in Nature Methods. Excited to see what the community discovers with this resource mapping ~15,000 human genes!

rdcu.be/ezGre

Big shout out to the whole team: Alán F. Muñoz, Tim Treis, @shatavishadg.bsky.social , @fabiantheis.bsky.social ntheis.bsky.social, @drannecarpenter.bsky.social, @shantanu-singh.cc

6/6

🔬API-first feature extraction for image-based profiling workflows

If you need to obtain interpretable features from your segmented microscopy images, but want to do it in a fully automated way, we know the struggle.

1/6

🌟 𝗥𝗲𝗺𝗶𝗻𝗱𝗲𝗿 🌟
The deadline for abstract submissions for oral presentations at Cytodata 2025 in Berlin is approaching!

👉 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗮𝗯𝘀𝘁𝗿𝗮𝗰𝘁 𝗯𝘆 𝗝𝘂𝗻𝗲 25!
cytodata25.eu-openscreen.eu/registration/

#BerlinConference #Imageanalysis #Microscopy

Delighted to see this out in print! It captures everything several of us in the field have been thinking about on the topic of measuring signal in high-dimensional profiling data, and I couldn't think of a better torchbearer and storyteller than @alxndrkalinin.bsky.social to champion this work.

A versatile information retrieval framework for evaluating profile strength and similarity - Nature Communications Profiling assays measure thousands of features to uncover biological insights but lack reliable methods for quality evaluation. Here, the authors develop a versatile information retrieval framework to...

🚨 New paper alert! We developed a versatile information retrieval framework that uses mean average precision (mAP) to robustly quantify sample activity and similarity in large-scale profiling data. Now out ‪@natcomms.nature.com: doi.org/10.1038/s414...

More in the 🧵 below:
1/7

Aleatoric and epistemic uncertainty are clear-cut concepts, right? ... right? 😵‍💫 In our new ICLR blogpost we let different schools of thought speak and contradict each other, and revisit chatbots where “the character of aleatory ‘transforms’ into epistemic” iclr-blogposts.github.io/2025/blog/re...

🚀🔬🦠 Releasing 🤖Cellpose-SAM🤖, a cellular segmentation algorithm with superhuman generalization 🦸‍♀️. Try it now on 🤗 huggingface.co/spaces/mouse...

paper: www.biorxiv.org/content/10.1...
w/ @computingnature.bsky.social 1/n

"Ich askid ChatGPT," Well Ich askid the stones, and the forest, and the rayne, and the wynde, and what thei seyde was learninge, and dreames, and growinge thinges, and a worlde wher we talke to each othir.

Model to Meaning: How to interpret statistical models with marginaleffects for R and Python

📚😅🎉

Yay!! I just submitted the complete manuscript of my upcoming book to the publisher!

Learn to easily and clearly interpret (almost) any stats model w/ R or Python. Simple ideas, consistent workflow, powerful tools, detailed case studies.

Read it for free @ marginaleffects.com

#RStats #PyData

🚨Latin American Workshop Series🚨

The team from @bethcimini.bsky.social‘s lab at @broadinstitute.org and the Center for Open Bioimage Analysis will host three free online image analysis workshops for LATAM ... in SPANISH🇪🇸 and PORTUGUESE🇧🇷!!!🎉🎊

broad.io/latam_worksh...

Deadline: April 23rd, 2025

Ask Erin, Dear Beth On Ask Erin/Dear Beth, bioimage analysis experts Beth Cimini and Erin Weisbart, of the Imaging Platform at the Broad Institute of MIT and Harvard, answer your image analysis questions! Whether it’s ab...

Delighted to announce that @erinweisbart.bsky.social and I have teamed up to create a new bioimage analysis video podcast called Ask Erin/Dear Beth - you can check it out at the link below! It will highlight common challenges in #bioimageanalysis, as well as our favorite solutions to them. (1/x)

Software engineering intern - summer/fall 2025 - Cambridge, MA USA If you are a current student (undergraduate/masters/PhD) with permission to work in the US and ability to work in-person in Cambridge, MA, US, consider a summer internship in the Broad Institute Imagi...

We're once again hiring a summer (+?) #bioimage #bioimageanalysis #software intern! Due to requirements of the funding program, you must be a current student, as well as work onsite in MA (+ have US work permission). Details at the link below. Spend your summer making great tools with fun people!

A research lab from Northwestern, chosen because it's generic and sort of zoomed out. Three scientists are visible in lab coats, and there are benches and shelving, with glass along one wall showing another high-rise building nearby. Overhead fluorescents provide light.

This is a room where we turn very modest salaries and budgets (and lots of coffee) into new knowledge, life-saving innovations, and technology that feeds business growth.

It's literally the loom that spins hay into gold but these numpties are suddenly worried about the cost of hay.

A genome-wide atlas of human cell morphology - Nature Methods An optical pooled cell profiling platform (PERISCOPE) based on Cell Painting and optical sequencing of molecular barcodes was used to develop the first unbiased genome-wide morphology-based perturbati...

Our paper “A genome-wide atlas of human cell morphology” is finally out today in @naturemethods.bsky.social ! www.nature.com/articles/s41...

(I tweeted about our preprint in 2023 over at the bad place, but deactivated my account, so here we go again!)

yep, but that would have been too easy ;)

and I regenerated my ssh keys 5 times over, began doubting everything lol

🧪 Summer internship alert, Feb 7 deadline

New URL: hsph.harvard.edu/fellowship-s...

(+ @harvardchanschool.bsky.social is now on 🦋!)

Hey #StatsSky, what are you favorite papers to cite when you need to justify something that is obvious (I once had a reviewer ask we justify the use of logistic regression on a binary outcome)
or when you need to push-back on silly reviewer requests (e.g., asking for p-values in table 1)?

I don't think it's widely clear to the #RadiologyAI community just how poorly GPT-4V compares with the top report generation models on chest X-rays, like MedVersa or MAIRA-2.

It's clear we need a way to track progress.

and @heidirehm.bsky.social too 🚀

@rajpurkar.bsky.social is here!

Instead of listing my publications, as the year draws to an end, I want to shine the spotlight on the commonplace assumption that productivity must always increase. Good research is disruptive and thinking time is central to high quality scholarship and necessary for disruptive research.

Over the long term there will be progress in closing the "capability-reliability gap" for agents, but for now, I think successful applications will be ones where (1) the user is in the loop, (2) errors are relatively easy to spot and (3) aren't a deal-breaker if not spotted.

CytoSummaryNet is a Deep Sets-based approach that uses self-supervised contrastive learning in a multiple-instance learning framework. Try it out!

Paper: doi.org/10.1371/jour...

Code: github.com/carpenter-si...

With @johnarevalo.bsky.social @drannecarpenter.bsky.social Mehrtash Babadi

3/n

Robert van Dijk (www.linkedin.com/in/robert-v-...) developed CytoSummaryNet – a simple strategy to learn an optimal way to aggregate single-cell features into population-level profiles, outperforming traditional averaging on tasks like mechanism-of-action prediction. 2/n

(a) Human U2OS cells treated with dimethyl sulfoxide (DMSO) and stained using the Cell Painting assay, which employs six dyes in five channels to label eight cellular compartments. The top row (from left to right) shows mitochondrial staining; actin, Golgi, and plasma membrane staining; and nucleolar and cytoplasmic RNA staining. The bottom row (from left to right) displays endoplasmic reticulum staining, DNA staining, and a montage of all five channels (from Cimini et al. [21]). (b) Thousands of features are extracted from each segmented cell in microscopy images of wells. A learned function f(x) (CytoSummaryNet) aggregates this data into a single feature vector: the sample’s profile. (c) An in-depth look at the model architecture used in this study. The model consists of three elements: a function φ(x), which maps the input data from ℝD to ℝL space, a summation, which collapses the cell dimension, and ρ(z), which maps the collapsed representation from ℝN to ℝL space. (d) During training, replicate compound profiles are forced to attract each other (green arrows) and simultaneously repel every other compound (red arrows) in the learned feature space. Here, all forces are drawn for a single profile of compound B.

Taking pictures of cells with a microscope, then extracting thousands of features from them is uncannily effective for quantifying cell state, esp. for genes and chemicals (e.g., Cell Painting). But we often average the rich single-cell data to simplify analysis. Can we do better?
#bioML 🧪
1/n

True luxury is found in the simplest moments.