Mapping Gene Regulatory Dependencies with Single-Cell Resolution

on YouTube ↗27m

" summary

Population-scale pooled iPSC single-cell RNA-seq generates ~1M cells from 215 donors, 10 pools of ~20 lines, sampled at days 11/30/52, and maps cell-subtype-specific eQTLs by pseudo-bulking and a covariance-driven low-rank model that estimates per-cell effect sizes while accounting for repeat structure and unequal cell counts. The method outperforms discrete clustering for interaction discovery, finds colocalizations (schizophrenia, insomnia) only in precise subpopulations, and extends to perturbation screens (∼7,000 CRISPRi guides across ~26–34 lines) to start inferring causal regulatory graphs.

AI-written summary · about

" chapters (16)

1. 0:00Intro
2. 1:30Technological advances drive computational innovations
3. 2:57Connecting GWAS variants to molecular traits
4. 7:29Population-scale single-cell sequencing across neuronal differentiation
5. 13:48Power compared to single-variable interaction tests
6. 14:46Benefits compared to discretisation
7. 15:41Assessing continuous scale genetic effects on gene expression
8. 16:45Fine-mapping cellular states associated with human disease variants
9. 17:10Summary and outlook
10. 18:40Natural genetic variation meets CRISPRi perturbations
11. 19:43Determinants of the similarly of downstream effect profiles of individual targets
12. 20:21Cell line is a major driver of variation in the data
13. 20:44Independent causal Mechanisms principle
14. 21:14The independent causal mechanism principle in biology
15. 21:33Adapting the ICM principle to interventional single-cell data
16. 23:38Does pooling affect neuronal differentiation efficiency?