Neville Sanjana
@nevillesanjana.bsky.social
Scientist at the New York Genome Center & NYU.
http://sanjanalab.org
http://sanjanalab.org
If you’d like to learn more, please check out all the details in the CRISPore-seq preprint here:
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
Transcriptome-wide profiling of alternative splicing regulators with CRISPore-seq
Alternative splicing creates diverse RNA isoforms from individual genes, yet single-cell CRISPR screens are limited to gene-level quantification and cannot detect changes in alternative splicing and t...
www.biorxiv.org
November 27, 2025 at 11:59 AM
If you’d like to learn more, please check out all the details in the CRISPore-seq preprint here:
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
Also, our great collaborators at @nanoporetech.com (@sisseljuul.bsky.social #DavidDai & #PriyeshRughani) have been a pleasure to work with on this and other long-read sequencing projects that I hope to tell you about soon. And thanks also to co-authors #WellsBurrell and #ZharkoDaniloski too.
November 27, 2025 at 11:59 AM
Also, our great collaborators at @nanoporetech.com (@sisseljuul.bsky.social #DavidDai & #PriyeshRughani) have been a pleasure to work with on this and other long-read sequencing projects that I hope to tell you about soon. And thanks also to co-authors #WellsBurrell and #ZharkoDaniloski too.
In particular, grateful to Akash for launching the project and Simon for seeing it through to the end and adding some wonderful mechanistic work.
November 27, 2025 at 11:59 AM
In particular, grateful to Akash for launching the project and Simon for seeing it through to the end and adding some wonderful mechanistic work.
CRISPore-seq would not be possible without a great team. This work was led by talented co-first authors (#SimonMuller, @nathanaelandrews.bsky.social, @rachelyan.bsky.social, #AkashSookdeo) — a set of postdocs & PhD/MD-PhD students from our group.
November 27, 2025 at 11:59 AM
CRISPore-seq would not be possible without a great team. This work was led by talented co-first authors (#SimonMuller, @nathanaelandrews.bsky.social, @rachelyan.bsky.social, #AkashSookdeo) — a set of postdocs & PhD/MD-PhD students from our group.
... and also for training new AI-based virtual cell and DNA language models.
November 27, 2025 at 11:59 AM
... and also for training new AI-based virtual cell and DNA language models.
With emphasis recently in academia & biotech on generating massive-scale perturbation single-cell data (e.g. @recursionpharma.bsky.social, #TahoeTx, #XairaTx, and others), full transcript resolution will enable more detailed phenotyping for mechanistic biology...
November 27, 2025 at 11:59 AM
With emphasis recently in academia & biotech on generating massive-scale perturbation single-cell data (e.g. @recursionpharma.bsky.social, #TahoeTx, #XairaTx, and others), full transcript resolution will enable more detailed phenotyping for mechanistic biology...
Overall, CRISPore-seq is exciting given its potential to resolve isoforms for next-generation Perturb-seq.
November 27, 2025 at 11:59 AM
Overall, CRISPore-seq is exciting given its potential to resolve isoforms for next-generation Perturb-seq.
We also confirmed this with a functional rescue experiment: After CCND1 knockdown, only the isoform containing exon 2 is able to rescue growth (in 2D cell culture and 3D spheroids) and cell cycle progression.
November 27, 2025 at 11:59 AM
We also confirmed this with a functional rescue experiment: After CCND1 knockdown, only the isoform containing exon 2 is able to rescue growth (in 2D cell culture and 3D spheroids) and cell cycle progression.
Experimentally, we found that exon 2 of CCND1 is required for binding (via co-IP) of CDK6.
November 27, 2025 at 11:59 AM
Experimentally, we found that exon 2 of CCND1 is required for binding (via co-IP) of CDK6.
The predicted protein structure using AlphaFold3 shows that exon 2 of CCND1 is important for complex formation with CDK6, a kinase required for cell-cycle progression (G1/S). Exon 2 loss leads to a 2.4-fold decrease in the predicted CCND1/CDK6 protein interface.
November 27, 2025 at 11:59 AM
The predicted protein structure using AlphaFold3 shows that exon 2 of CCND1 is important for complex formation with CDK6, a kinase required for cell-cycle progression (G1/S). Exon 2 loss leads to a 2.4-fold decrease in the predicted CCND1/CDK6 protein interface.
One example where we took a deep dive is cyclin D1 (CCND1). Knockdown of the splicing factor SF3B4 doesn’t change overall expression of CCND1 but it changes the transcript (more skipping of exon 2).
November 27, 2025 at 11:59 AM
One example where we took a deep dive is cyclin D1 (CCND1). Knockdown of the splicing factor SF3B4 doesn’t change overall expression of CCND1 but it changes the transcript (more skipping of exon 2).
So, we combined CRISPore-seq with isoform-specific RNA-targeting CRISPR knockdown data. Here’s how integrating these datasets looks: We see many cases of specific exons that are excluded ✖️❌✖️but are part of essential transcripts‼️.
November 27, 2025 at 11:59 AM
So, we combined CRISPore-seq with isoform-specific RNA-targeting CRISPR knockdown data. Here’s how integrating these datasets looks: We see many cases of specific exons that are excluded ✖️❌✖️but are part of essential transcripts‼️.
Given the predominance of exon skipping in CRISPore-seq RBP perturbations, we wanted to understand whether we might find specific transcript isoforms that are important for cell fitness and survival that are being skipped/spliced out.
November 27, 2025 at 11:59 AM
Given the predominance of exon skipping in CRISPore-seq RBP perturbations, we wanted to understand whether we might find specific transcript isoforms that are important for cell fitness and survival that are being skipped/spliced out.
And, for those RBP’s with binding data (eCLIP from @geneyeo.bsky.social, @darnelr.bsky.social, & #ENCODE), we can see changes in splicing after RBP loss precisely at exons where the RBP binds.
November 27, 2025 at 11:59 AM
And, for those RBP’s with binding data (eCLIP from @geneyeo.bsky.social, @darnelr.bsky.social, & #ENCODE), we can see changes in splicing after RBP loss precisely at exons where the RBP binds.
What’s most exciting is that we can deeply catalog 📕📗📘📙📚alternative splicing events after RBP perturbation using the long reads from CRISPore-seq.
We found that exon skipping was the most frequent splicing alteration in our dataset of RBP perturbations. 🦘🦘🦘
We found that exon skipping was the most frequent splicing alteration in our dataset of RBP perturbations. 🦘🦘🦘
November 27, 2025 at 11:59 AM
What’s most exciting is that we can deeply catalog 📕📗📘📙📚alternative splicing events after RBP perturbation using the long reads from CRISPore-seq.
We found that exon skipping was the most frequent splicing alteration in our dataset of RBP perturbations. 🦘🦘🦘
We found that exon skipping was the most frequent splicing alteration in our dataset of RBP perturbations. 🦘🦘🦘
The RBP knock-downs result in a clear set of differentially-expressed transcripts and there is a correlation between how essential a RBP is and how many transcripts are disrupted.
November 27, 2025 at 11:59 AM
The RBP knock-downs result in a clear set of differentially-expressed transcripts and there is a correlation between how essential a RBP is and how many transcripts are disrupted.
Using CRISPore-seq, we decided to perturb several RNA-binding proteins (RBPs) — with different functions and ones that are more or less essential — and map their impact across the long-read transcriptome.
November 27, 2025 at 11:59 AM
Using CRISPore-seq, we decided to perturb several RNA-binding proteins (RBPs) — with different functions and ones that are more or less essential — and map their impact across the long-read transcriptome.
And this is even clearer when we look at individual transcripts and how their exon connectivity changes with different genetic perturbations.
We found that long reads resolved 85% of isoforms, whereas short reads could distinguish only 20%.
We found that long reads resolved 85% of isoforms, whereas short reads could distinguish only 20%.
November 27, 2025 at 11:59 AM
And this is even clearer when we look at individual transcripts and how their exon connectivity changes with different genetic perturbations.
We found that long reads resolved 85% of isoforms, whereas short reads could distinguish only 20%.
We found that long reads resolved 85% of isoforms, whereas short reads could distinguish only 20%.
We detect 1000s of transcripts in single-cells that would not be possible with short-read — even with many less UMIs per cell.
November 27, 2025 at 11:40 AM
We detect 1000s of transcripts in single-cells that would not be possible with short-read — even with many less UMIs per cell.
So, to answer the big question, can we detect more of the diverse isoforms in the human transcriptome?
💯 YES! 🚀🚀🚀
With CRISPore-seq, there is uniform coverage of transcripts from the 5’ end to the 3’ end.
💯 YES! 🚀🚀🚀
With CRISPore-seq, there is uniform coverage of transcripts from the 5’ end to the 3’ end.
November 27, 2025 at 11:40 AM
So, to answer the big question, can we detect more of the diverse isoforms in the human transcriptome?
💯 YES! 🚀🚀🚀
With CRISPore-seq, there is uniform coverage of transcripts from the 5’ end to the 3’ end.
💯 YES! 🚀🚀🚀
With CRISPore-seq, there is uniform coverage of transcripts from the 5’ end to the 3’ end.
Original ECCITE-seq study (w/ @psmibert.bsky.social): sanjanalab.org/reprints/Mim...
COVID-19 host genes: sanjanalab.org/reprints/Dan...
Noncoding variant discovery: sanjanalab.org/reprints/Mor...
COVID-19 host genes: sanjanalab.org/reprints/Dan...
Noncoding variant discovery: sanjanalab.org/reprints/Mor...
November 27, 2025 at 11:40 AM
Original ECCITE-seq study (w/ @psmibert.bsky.social): sanjanalab.org/reprints/Mim...
COVID-19 host genes: sanjanalab.org/reprints/Dan...
Noncoding variant discovery: sanjanalab.org/reprints/Mor...
COVID-19 host genes: sanjanalab.org/reprints/Dan...
Noncoding variant discovery: sanjanalab.org/reprints/Mor...
For CRISPore-seq, we used the same 5’ capture strategy (@10xgenomics.bsky.social) as in ECCITE-seq to combine direct CRISPR guide RNA capture with long AND short read sequencing.
November 27, 2025 at 11:40 AM
For CRISPore-seq, we used the same 5’ capture strategy (@10xgenomics.bsky.social) as in ECCITE-seq to combine direct CRISPR guide RNA capture with long AND short read sequencing.
To get a transcript-level view of the effect of CRISPR perturbations, we teamed up w/ @sisseljuul.bsky.social & friends @nanoporetech.com to add long-read sequencing to Perturb-seq:
November 27, 2025 at 11:40 AM
To get a transcript-level view of the effect of CRISPR perturbations, we teamed up w/ @sisseljuul.bsky.social & friends @nanoporetech.com to add long-read sequencing to Perturb-seq:
This look at the latest human genome annotation from GENCODE highlights the problem:
There are almost 10-fold more protein-coding transcripts (~200,000) than genes (~20,000).
There are almost 10-fold more protein-coding transcripts (~200,000) than genes (~20,000).
November 27, 2025 at 11:40 AM
This look at the latest human genome annotation from GENCODE highlights the problem:
There are almost 10-fold more protein-coding transcripts (~200,000) than genes (~20,000).
There are almost 10-fold more protein-coding transcripts (~200,000) than genes (~20,000).