Big thanks to Sam Regalado, @‪cxqiu.bsky.social‬, @jshendure.bsky.social and all co-authors for sharing this journey. Also check out the accompanying preprint applying barcoded monoclonals for lineage recording with @choijunhong.bsky.social. www.biorxiv.org/content/10.1...

Of note, mosaic screens as performed by us or others can still be valuable tools, especially if the extent of heterogeneity is controlled by culturing conditions or quantified up front in absence of perturbations and taken into consideration by sufficient sampling and through biological replicates.

This platform is versatile (EBs are precursors for various organoids), scalable, and opens up new applications, such as studying genetic diseases with reduced penetrance or variable expressivity.

With organoid barcodes we can quantify baseline inter-EB heterogeneity and perturbation-specific changes in cell type composition. We can also apply replicate-based DE analyses, avoiding the high false positive rate caused by inter-EB heterogeneity in standard single cell DE.

Our solution: monoclonal embryoids with an ‘organoid barcode’. This allows us to determine not only the transcriptome and perturbation identity for each cell but also the individual EB of origin - while still growing EBs in a pool, for scalability and shared culture conditions.

Many gRNAs seemed to have significant effects on cell type composition in mosaic EBs, but these were not reproducible across replicates: Stochastic clonal skewing of individual gRNAs in EBs, combined with inter-EB heterogeneity, overwhelms most perturbation effects and reduces statistical power.

Nice work, congrats Sud!