Refining our understanding of the evolutionary mechanisms of different types of cancer in order to achieve increasingly targeted diagnoses and therapies is the objective of the new RESOLVE method, based on a study of ‘mutational signatures’, i.e. recurrent patterns of mutations in DNA that tell the story of the damage suffered by cancer cells and help identify their origin and mechanisms of development. 

The study presents a new computational tool to analyse the mutational mechanisms underlying cancer. By applying this method to approximately 20,000 adult and paediatric cancer genomes, the researchers were able to accurately identify a small number of dominant mutational signatures, associated with both known biological mechanisms (such as aging, smoking exposure, or defects in DNA repair) and different clinical prognoses. 

‘The problem of identifying processes that generate mutations in DNA is crucial to understanding what damages the genome and can accelerate tumour transformation. The tools we create in our laboratories are able to extract these signals thanks to machine learning techniques. This area of research sees us directly involved with several projects, such as this one in collaboration with Milano-Bicocca and others that we coordinate within our group,’ says Giulio Caravagna, professor of Computer Science at UniTS involved in the study. 

Mutation signature analysis is an established practice in cancer genomics but presents several challenges. Compared to existing methods, RESOLVE (Robust EStimation Of mutationaL signatures Via rEgularization) allows a more precise detection of mutation signatures, a more reliable estimate of their relevance in individual patients and the possibility of distinguishing tumours into molecular subtypes, with promising repercussions for personalised medicine.

This innovative method, illustrated in the article ‘Comprehensive analysis of mutational processes across 20 000 adult and pediatric tumors’ published in the journal Nucleic Acids Research, was developed by a multidisciplinary group of the University of Milan-Bicocca, coordinated by Daniele Ramazzotti (Department of Medicine and Surgery and Foundation of the Research Hospital San Gerardo dei Tintori). Researchers from the Department of Computer Science Marco Antoniotti and Alex Graudenzi, from the Department of Medicine Rocco Piazza and Luca Mologni, and Giulio Caravagna from the University of Trieste also participated in the project. The team also includes Matteo Villa, Federica Malighetti, Luca De Sano, Alberto Maria Villa, Nicoletta Cordani and Andrea Aroldi.