Scalable genetic screening for regulatory circuits using compressed Perturb-seq
Level 5 - mechanism / opinion, no new human data
Level 5 by design analogy; in vitro functional genomics and computational methodology.
OpenAlex W4387880996 · doi:10.1038/s41587-023-01964-9
What was done
The authors developed compressed Perturb-seq, a functional genomics framework combining pooled CRISPR screens and single-cell RNA sequencing with compressed sensing algorithms. The method introduces multiple random perturbations per cell or captures multiple cells per droplet, computationally decompressing the output by leveraging the sparse architecture of gene regulatory networks. The approach was evaluated across 598 genes involved in the immune response to bacterial lipopolysaccharide.
What was found
The abstract reports no specific numerical concordance values, sequencing counts, or cell numbers. It reports that compressed Perturb-seq matched the accuracy of standard Perturb-seq with an order-of-magnitude cost reduction and increased power to detect genetic interactions across 598 genes. The screen identified known and novel immune regulators and found evolutionarily constrained genes with downstream targets enriched for immune disease heritability.
Why it matters
Pooled single-cell CRISPR screens are typically cost-prohibitive for large-scale combinatorial experiments. This compressed sensing approach enables substantial cost reductions and scalable mapping of genetic regulatory networks and disease-relevant pathways.
Limits
The abstract lacks quantitative metrics such as exact decompression error rates, sample sizes (cell counts), and statistical thresholds. The decompression algorithm relies on the assumption of network sparsity, which may not hold across all biological contexts or densely connected transcriptional programs.
Cited by
- supports Single-cell RNA sequencing combined with CRISPR perturbation allows simultaneous measurement of the delivered CRISPR guide and the full transcriptome state of individual primary human immune cells.