Genome assembly of two allotetraploid cotton germplasms reveals mechanisms of somatic embryogenesis and enables precise genome editing.

Somatic embryogenesis is crucial for plant genetic engineering, yet the underlying mechanisms in cotton remain poorly understood. Here we present a telomere-to-telomere assembly of Jin668 and a high-quality assembly of YZ1, two highly regenerative allotetraploid cotton germplasms. The completion of the Jin668 genome enables characterization of ~30.1 Mb of centromeric regions invaded by centromeric retrotransposon of maize and Tekay retrotransposons, an ~8.1 Mb 5S rDNA array containing 25,190 copies and a ~75.1 Mb major 45S rDNA array with 8,131 copies. Comparative analyses of regenerative and recalcitrant genotypes reveal dynamic transcriptional patterns and chromatin accessibility during the initial regeneration process. A hierarchical gene regulatory network identifies AGL15 as a contributor to regeneration. Additionally, we demonstrate that genetic variation affects sgRNA target sites, while the Jin668 genome assembly reduces the risk of off-target effects in CRISPR-based genome editing. Together, the complete Jin668 genome reveals the complexity of genomic regions and cotton regeneration, and improves the precision of genome editing.