Identifying strain-specific functional genes in colorectal cancer

Changes in the gut microbiota have been associated with colorectal cancer (CRC), but neither the causal mechanisms nor corresponding microbial strains and small molecule products have been elucidated for CRC. We have developed a new strain-level meta-analysis using stool metagenomic profiles of 600 CRC patients, 143 with precancerous adenomas, and 662 healthy controls from nine recently published CRC microbiome studies. We created the MMUPHin framework to jointly normalise these datasets and identify potential consistently significant links between CRC neoplasia, severity, and microbial species and strains. We identified several species as novel CRC biomarkers including several typical oral species. We observed that CRC cases were depleted in geographically-specific Prevotella copri subtype carriers. A group of functional genes unique to subsets of Escherichia coli strains was associated with CRC phenotypes, comprising annotations to transporters, type II secretion systems, flagellar and sulfur metabolism. This study adds further evidence to the hypothesis that strain-level genomic variation in gut microbes may be a major driver in the initiation or development of CRC.