From the left: José Córdoba, Marcos Díaz Gay and Pilar Gallego. Crédito: Laura M. Lombardía / CNIO
The new Digital Genomics Group will be led by Galician researcher Marcos Díaz Gay, arriving from the University of California, San Diego (USA)
They will develop tools based on AI and machine learning that accurately identify which mutation patterns – or mutational signatures – correlate with each tumor
Their goal is to use mutational signatures to improve tumor diagnosis, prognosis of tumor progression and response to potential treatments
Some tumors are caused by one or more inherited mutations in specific genes. Others are the result of the accumulation, over a lifetime, of mutations induced by environmental factors or lifestyle habits. Over time, research has identified some of these mutations as being directly responsible for causing a tumor. However, there are many other mutations that have gradually accumulated in tumors. As their relationship with cancer was not so evident, until a decade ago they were not taken into account in genetic studies searching for the origin of tumors.
Until “a more comprehensive perspective was adopted and all the mutations in a tumor began to be analyzed. The aim was to be able to detect patterns among them that might correlate with a particular type of tumor,” explains Marcos Díaz Gay, a specialist in the identification of these patterns. “This is additional information that was previously lost, even though it can explain a favorable context for a tumor to arise and also how the tumor will develop depending on the patient or what response it may have to different drugs,” he adds.
Genomic archaeology of a tumor
This Galician researcher has spent the last five years at Ludmil B. Alexandrov’s lab, one of the world leaders in the research of these patterns, called ‘mutational signatures’. He has now arrived at the Spanish National Cancer Research Center (CNIO) as head of the new Digital Genomics Group, which has already been joined by postdoctoral researchers Pilar Gallego and José Córdoba.
The CNIO Digital Genomics Group is part of the center’s artificial intelligence (AI) strategy launched with funding from the Ministry of Digital Transformation. In addition to creating new junior groups – led respectively by Díaz Gay and Roger Castells Graells – and initiating two specific research projects, the CNIO will strengthen AI in different groups.
Searching for mutational signatures is equivalent to doing “genomic archeology”; we try to trace the tumor history basing of the changes that occur in its genome, and thus we obtain information on which processes or exposure to certain factors have given rise to the tumor. For example, “smoking causes an accumulation of mutations that form such a characteristic pattern that, when we identify it in a lung tumor, we can know –without having to ask directly– whether the patient has been a smoker,” says the researcher.
Fine-tuning mutational signatures
To get there, it is necessary to sequence the DNA of both the tumor and other “healthy” tissue of the patient, which serves as a control. In this healthy tissue – commonly blood, in the case of solid tumors – the inherited mutations are identified. These are then compared with those present in the tumor, the inherited mutations are ‘subtracted’ and the remaining mutations are identified as exclusive to the tumor, which are known as somatic mutations.
This comparison and the identification of patterns require specific bioinformatics tools. “Developing them is my specialization and the great contribution that my group can make to the CNIO,” says Díaz Gay. His goal is to “give a boost to this methodology to try to improve it, and to be able to use mutational signatures to diagnose tumors, predict the evolution of the disease, evaluate the most appropriate treatment or predict the response to different therapeutic options”.
In parallel, they are already working to identify whether mutation patterns vary in different populations, be it due to hereditary factors, genetic ancestry, or environmental exposure to certain mutagens.
Artificial intelligence and computational capacity
It is worth pointing out that the identification of mutation patterns is based on artificial intelligence and machine learning, and requires enormous computational capacity. “Identifying high-definition mutational patterns can take days or weeks of computation on a computer” clarifies the computational biology specialist. He considers that the incorporation of his group is part of the effort being made by the CNIO, with the economic support of the Ministry of Digital Transformation, to improve this capacity “because we can develop new computational tools and gain access to genomic databases and supercomputers”.
Similarly, he values having the CNIO’s infrastructure at his disposal, such as the biobank and the next-generation sequencer that the center acquired last year, “since we would also like to recruit our own cases, for example, to study lung cancer in non-smokers”.
Marcos Díaz Gay does not hide his satisfaction at having returned to his country – “we all like to return home”–, but, when deciding about his coming back, this engineer by training valued the opportunity to contribute “to the growth of computational biology in a center of excellence in cancer research such as the CNIO, with the synergies offered for our line of work by other groups, both those already active in bioinformatics and genomics, and those that are joining now, thanks to the new commitment to integrate artificial intelligence in cancer research”.