Deconstructing evolutionary histories of complex genomic rearrangements in lung malignancies

Somatic genomic rearrangements are hallmarks of cancer. Complex genomic rearrangements (CGRs) involving multiple intertwined structural alterations are often present in tumor genomes. CGRs frequently harbor oncogenic drivers, but their genomic architectures and etiologies are poorly understood. We used deep-coverage optical mapping technology to profile the genomic landscapes of normal lung tissues, benign pulmonary lesions, carcinoma in situ, and advanced carcinomas to examine the patterns of genome disorganization and instability in different stages of carcinogenesis in lung. Large rearrangements and CGRs were prevalent in the carcinomas. We developed omcplR to resolve the architecture of CGRs and predict their genesis from optical mapping data using genome-graph concept. We found that CGRs often arose from hierarchical combinations of multiple, localized simple structural variations, and harbored allelic heterogeneity at the affected loci. Rearrangement patterns and associated genomic signatures suggested that chromoanasynthesis was a likely prevalent mechanism driving complex genomic rearrangements. The early rearrangement junctions in intra-chromosomal CGRs were usually localized within the same chromatin domains, but the late junctions in advanced tumors had more heterogeneous contexts suggesting progressive organizational heterogeneity. The CGRs, especially the late events therein were under positive selection. A composite signature of genomic alterations including the CGRs captured the trajectory of progressive genomic disorganization and instability with carcinogenesis in lung and underscored the extent of genomic structural heterogeneity among the in-situ tumors.