New class of cancer mutations discovered in so-called ‘junk’ DNA with the help of machine learning

Good news! Cancer is history (soon)! Bring the power of machine learning & AI to bear!

"Non-coding DNA – the 98% of our genome that doesn’t contain instructions for making proteins – could hold the key to a new approach for diagnosing and treating cancers ...
The findings ... reveal mutations in previously overlooked regions of the genome that may contribute to the formation and progression of at least 12 different cancers, including prostate, breast and colorectal."

From the abstract:

"CCCTC-binding factor (CTCF) is an insulator protein that binds to a highly conserved DNA motif and facilitates regulation of three-dimensional (3D) nuclear architecture and transcription. CTCF binding sites (CTCF-BSs) reside in non-coding DNA and are frequently mutated in cancer. Our previous study identified a small subclass of CTCF-BSs that are resistant to CTCF knock down, termed persistent CTCF binding sites (P-CTCF-BSs). P-CTCF-BSs show high binding conservation and potentially regulate cell-type constitutive 3D chromatin architecture. Here, using ICGC sequencing data we made the striking observation that P-CTCF-BSs display a highly elevated mutation rate in breast and prostate cancer when compared to all CTCF-BSs. To address whether P-CTCF-BS mutations are also enriched in other cell-types, we developed CTCF-INSITE—a tool utilising machine learning to predict persistence based on genetic and epigenetic features of experimentally-determined P-CTCF-BSs. Notably, predicted P-CTCF-BSs also show a significantly elevated mutational burden in all 12 cancer-types tested. Enrichment was even stronger for P-CTCF-BS mutations with predicted functional impact to CTCF binding and chromatin looping. Using in vitro binding assays we validated that P-CTCF-BS cancer mutations, predicted to be disruptive, indeed reduced CTCF binding. Together this study reveals a new subclass of cancer specific CTCF-BS DNA mutations and provides insights into their importance in genome organization in a pan-cancer setting."

New class of cancer mutations discovered in so-called ‘junk’ DNA | Garvan Institute of Medical Research Using artificial intelligence, Garvan Institute researchers have found potential cancer drivers hidden in so-called ‘junk’ regions of DNA, opening up possibilities for a new approach to diagnosis and treatment.

Machine learning enables pan-cancer identification of mutational hotspots at persistent CTCF binding sites (open access)

Graphical abstract