PTTM: dissecting the profile of tumor tissue microbiome to reveal microbiota features and associations with host transcriptome

Microbiota is present in the human tissue microenvironment and closely related to tumorigenesis and treatment. However, the landscape of tissue microbiome and its relationship with tumors remain less understood. In this study, we re-analyzed the omics data from the 7104 samples (94 projects for 15 cancers) in the NCBI database to obtain microbial profiles. After normalization and decontamination processing, we established classification models to distinguish between different tumors and tumor with adjacent normal tissues. The models had excellent performances, indicating that tissue microbiome had significant tumor specificity. Moreover, a series of key bacteria and bacteria-gene association pairs were screened out based on bioinformatic analysis, such as the tumor-promoting bacteria Fusobacterium, the tumor-suppressing bacteria Actinomyces, and the significant Rhodopseudomonas-COL1A1 association pair. In addition, we created a visual website, PTTM (http://198.46.152.196:7080/), for users to query and download the results. The identified key bacteria and association pairs provide candidate targets for further exploration of the molecular mechanisms of microbial action on tumorigenesis and the development of cancer therapy.