Prediction of the 3D cancer genome from genomic rearrangements using InfoHiC

Abstract

Although cancer genomes often contain complex genomic rearrangements, its impact on tumorigenesis is still unclear, especially when they are involved in non-coding regions. Understanding 3D genome architecture is crucial for uncovering the impacts of genomic rearrangements. Here, we present InfoHiC, a method for predicting 3D genome folding and cancer Hi-C from complex genomic rearrangements. InfoHiC provides distinct interaction views of multiple contigs from the cancer Hi-C matrix. We then validated cancer Hi-C prediction using breast cancer cell line data and found contig-specific interaction changes. Moreover, we applied InfoHiC to patients with breast cancer and identified neo topologically associating domains and super-enhancer hijacking events associated with oncogenic overexpression and poor survival outcomes. Finally, we applied InfoHiC to pediatric patients with medulloblastoma, and found genomic rearrangements in non-coding regions that caused super-enhancer hijacking events of medulloblastoma driver genes (GFI1, GFI1B, and PRDM6). In summary, InfoHiC can predict genome folding changes in cancer genomes and may reveal therapeutic targets by uncovering the functional impacts of non-coding genomic rearrangements.