Leads: Matt Suderman and Paul Yousefi

Cancer develops when cells stop performing their normal functions and redirect energy and other resources toward ensuring their own survival and replication. Successful treatment depends on detecting the presence of cancer before it has spread through the body.

Fortunately, many types of cancer cells can be distinguished from healthy cells by their abnormal behaviours, their genomes or their molecular by-products. Our aim is to systematically investigate these abnormalities in order to devise tests for the early detection of cancer. Toward this aim, we apply state-of-the-art machine-learning methods to multi-omic datasets derived from tissues collected from individuals with a variety of cancers. We focus on peripheral tissues such as blood and saliva that can be collected using minimally invasive techniques. Omics of interest include the genome, proteome, methylome, metabolome and microbiome.

Our data sources include well-known studies such as ALSPAC, EPIC, EPIC-Italy, GODMC, Generation Scotland, Head & Neck 5000, HUNT, LC3, MCCS, NOWAC, NFBC 1966, ProtecT, UK Biobank and Understanding Society.