Flyer says: Stand up for Science 2025. March 7, 2025 Washington, DC and nationwide. Science is for Everyone. Find your local rally site and other ways to get involved at standupforscience2025.org
I will be there!
(note: the speaking portion of the event has been moved to Boston Common)
#standupforscience2025
www.eventbrite.com/e/stand-up-f...
If the block on the Federal Register isn’t removed in the next few months, I fear a majority of labs will close within a year. This small procedural wrench has the power to kill US science indefinitely, losing an entire generation of discovery and innovation.
I don't know who needs to hear this, but none of the preemptive obedience is going to save us.
Thanks Gary - I thought I saw you there!
Side note: if you happen to be attending #SLAS2025, I’ll be talking about this work at 10AM Tuesday in the Functional Genomics session, stop by and say hi! <fin>
All of our data and analysis tools are open source and freely available (see Code and Data availability), and there is much more in the paper than I was able to cover here, so check it out and let us know what you think!
We think this atlas will be a useful resource for both biological interrogation and for the development and testing of new image analysis methods.
In addition, we present a draft genome scale atlas of human cell morphology containing more than 30 million perturbation-assigned cell images and high-dimensional single cell profiles.
We believe this combination of accessibility and cost effectiveness make PERISCOPE-style screens a democratizing platform technology for linking genotypes to cellular programs.
In sum, pooled optical screens are a powerful approach for generating high-dimensional genotype-phenotype maps, and we demonstrate that these maps can now be generated routinely (and at a low cost per cell profile) at genome scale using standard imaging equipment and open source analysis pipelines.
As an aside, though this finding was scooped by two concurrent papers in @science.org while we were preparing the preprint, it’s encouraging to see a hit from our unbiased imaging screen robustly validated by numerous groups!(www.science.org/doi/10.1126/..., www.science.org/doi/10.1126/...)
figure showing genes that had high cosine similarity to TMEM251 in our screen
We were also able to use this clustering-by-morphological similarity information (plus mechanistic follow-up) to identify a role for the poorly characterized gene TMEM251 in lysosomal protein trafficking through the M6P-system.
network diagram showing HeLa DMEM and HPLM screen hit cellular processes
as well as identifying shared and media-specific responses to gene perturbation in cells cultured in DMEM vs human plasma-like media (HPLM).
UMAP showing clustering of genes by physiological function from the HeLa DMEM screen
These screens generated rich phenotypic data, enabling us to use morphological profile similarity to explore relationships between perturbed genes such as clustering protein complex members by morphological similarity, and capturing membership/directionality of signaling pathways.
We then applied these methods to execute 3 whole genome (80,862 sgRNAs targeting 20,393 genes) CRISPR KO screens in A549, and HeLa (paired screens with different growth media).
schematic showing the overall analysis pipeline
We ( @erinweisbart.bsky.social , @bethcimini.bsky.social , Greg Way) also built scalable, distributed open source analysis pipelines for alignment/cropping/background correction/barcode calling/etc. allowing us to turn millions of fluorescence images into sgRNA-assigned single cell profiles
PERISCOPE destaining schematic showing 5 color imaging followed by 4 color in situ sequencing-by-synthesis
To do this, we first built a variant of the Cell Painting panel that allowed destaining of fluorophores from a subset of markers (via cleavable disulfide or azidomethyl linker) so that we could perform four color in situ sequencing-by-synthesis without fluorescence overlap.
PERISCOPE logo PERISCOPE Acronym: Perturbation Effect Readout In situ via Single-cell optical phenotyping (yeah it's a mouthful)
So, we teamed up with Anne and Paul’s lab, along with Calvin Jan at Calico, with the aim of building an accessible & unbiased high-content cell profiling platform that could be applied to genome-scale CRISPR screens (Project PERISCOPE).
Fortunately, the Blainey Lab and others solved this problem by developing methods to sequence individual perturbation barcodes using in situ sequencing, so we can now assign quantitative cell phenotypes AND barcodes to individual cells, enabling optical pooled screens. www.cell.com/cell/fulltex...
Such data can be generated easily & cheaply, but historically hasn’t been ideal for large-scale functional genomics because imaging experiments could not use pooled genetic perturbation libraries (no way to optically determine which cell received which perturbation).
Cell images, especially multicolor fluorescence images, are an immense source of phenotypic information, and thanks to approaches like Cell Painting from @drannecarpenter.bsky.social , we can extract this information as quantitative features (size, shape, texture, etc.) in automated fashion.
Even so, and despite amazing recent advances in this space, such as www.cell.com/cell/fulltex... from @weissmanlab.bsky.social, these studies can still be quite difficult to execute at genome-scale. So we turned to another approach: optical fluorescence imaging.
We’re big fans of using expression-based profiling (specifically Perturb-seq) to generate high-dimensional phenotypic profiles of perturbed cells, and w/ @oana-ursu.bsky.social & @boehmjesse.bsky.social have used it successfully for generating variant effect maps (www.nature.com/articles/s41...)
For the last several years, a major focus of our lab has been building scalable tools to allow us to link genotypes to cellular programs at genome scale to understand gene and variant function in cancer and metabolic diseases.
First - this was a bi-coastal, multi-lab project w/ many folks to recognize including fellow blueskyers: @erinweisbart.bsky.social @bethcimini.bsky.social @shantanu-singh.cc @juliabauman.bsky.social @skavari.bsky.social @drannecarpenter.bsky.social
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!)
toddler is all cleaned up and settled on the plane with her headphones on.
now she’s exclaiming “I’m on a call!” every time there’s an inflight announcement
“in an Uber” is not the best place to discover that your toddler gets carsick
