Chromosomal aberrations (CAs) are a hallmark of genomic instability, playing a critical role in the evolution of cancer. While extensive research has linked CAs to mitotic errors and nuclear atypias, the underlying processes and spontaneous formation rates of CAs in human cells have remained insufficiently explored. In this talk, I will present our group’s advances in understanding de novo CA formation through coupling automated live-cell imaging and single-cell genomics, termed MAGIC (Machine Learning-Assisted Genomics-and-Imaging Convergence). MAGIC autonomously integrates real-time imaging of micronucleated cells, machine learning, and genomics, allowing for high-throughput investigation of CA events across successive cell cycles in near-diploid, non-transformed cell lines through single cell genomic sequencing at scale. Our findings reveal dicentric chromosomes as a frequent initiating event and establish a baseline CA rate, which notably increases in TP53-deficient cells. Furthermore, we find that DNA double-strand breaks induce distinct CA processes, leading to diverse and complex outcomes, depending on the chromosomal location of the initiating DNA break.

In addition, I will discuss the functional impact of mosaic structural variants (SVs) in normal tissues, particularly within hematopoietic stem and progenitor cells. Using Strand-seq, we sequenced over a thousand single-cell genomes from human donors of varying ages, uncovering cell type-specific SV mosaicisms landscapes that become more pronounced with age and contribute to cellular dysfunction by disrupting key molecular pathways.