Deep Learning Predicts Nuclear Lamina-Associated Chromatin

Summary

LADDER is a deep learning framework designed for predicting tethering of chromatin to the nuclear lamina via DNA sequence and associated genomic properties. By combining it with genome-wide random in silico deletions, the authors revealed an intrinsic genome lamina susceptibility landscape. LADDER can learn general features of lamina organization from one species and use them to extrapolate predictions to other mammalian species. Leveraging this cross-species capability, the authors examined allele-specific differential lamina tethering associated with mouse escapee genes.