Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/202616/n It worked! For example, we recovered an allelic deletion series where some deletions overlap RBM3 while others spare it. In single-cell transcriptomes, only the RBM3-overlapping deletions reduced RBM3 expression, giving a clean genotype-phenotype link 110
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/202615/n Making this work required several tweaks. For example, in situ T7 transcripts are not naturally polyadenylated, so we adapted the workflow to polyadenylate T7-derived transcripts and capture them alongside endogenous mRNAs. Details are in the manuscript 100
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/202613/n Coming back to the original question: How much of the human genome is dispensable for the growth of a haploid cell line? Somewhere between 50-96%! Running Shred-seq at higher coverage should further narrow that empirical bound 111
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/202612/n For example, surviving deletions at this locus are depleted of deletions that intersect WDR3, an essential gene. Generally, Shred-seq lets us identify constrained loci by looking for places where the surviving deletion profile deviates from the expected Cas3 distribution 100
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/202611/n Beyond global trends, deletion scans allow us to look at locus-specific essentiality. If unselected deletions follow a baseline length distribution, then post-selection distortions of that distribution can reveal regions under constraint without relying on prior annotations 110
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/202610/n Is this all just avoidance of essential coding genes? Not entirely. Even when we restricted the analysis to non-coding deletions, surviving deletions were depleted from active chromatin and enriched in inactive chromatin features 200
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/20269/n Surviving deletions are depleted not only for essential protein-coding genes, but also for active, conserved, and mutation-constrained sequences. We even see a signal for lncRNAs (using a dataset generated by @nevillesanjana.bsky.social lab) 110
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/20267/n After lots of optimizations, we managed to make this really high throughput (gigabase-scale): We induced > 36k deletions between 100bp-500 kb that sum up to 2.55 Gb, and scanned ~14% of the human genome 110
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/20266/n With all the components in place, here is how Shred-seq works: We randomly integrate deletion launch sites ‘beacons’ into the genome, target Cas3 to them, make deletions, and read out the variants using phage promoters on the beacon 100
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/20264/n There is a CRISPR system that does exactly that! CRISPR-Cas3 can be targeted to any DNA sequence using a crRNA. Once there, it starts chewing up DNA, leaving large deletions in its wake! Fantastic pioneering work by the Yan Zhang lab 100
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/20263/n But not just any deletions. Ideally, we wanted deletion scans: many overlapping deletions of different lengths starting from defined sites. Such an allelic series should make it possible to identify where essential genomic regions begin and end with high resolution 100
Jonas Koeppel @jonaskoeppel.bsky.social · 04/06/20261/n What fraction of the human genome is essential for cells? Excited to share our preprint that explores this question by combining an unusual CRISPR system, phage promoters, and thousands of deletion launchpads. @sudpinglay.bsky.social @jshendure.bsky.social www.biorxiv.org/content/10.6... 28134