Identifying The Landscape Of Intratumoral Microbes Via A Single Cell Transcriptomic Analysis.

Microbial taxa that are differentially abundant between cell types are likely to be intracellular. Here we describe a new computational pipeline called CSI-Microbes (computational identification of ell type pecific ntracellular ) that aims to identify such putative intracellular species from single cell RNA-seq data in a given tumor sample. CSI-microbes also includes additional steps that can be applied to filter out microbial contaminants from the microbial residents of cells in the patients. We first test and validate CSI-microbes on a dataset of immune cells deliberately infected with . We then apply CSI-microbes to identify intracellular microbes in breast cancer and melanoma. We identify as differentially abundant in the tumor cells of one breast cancer sample. We further identify three bacterial genera and four fungal genera that are differentially abundant and hence likely to be intracellular in the tumor cells in melanoma samples. No cell type specific bacteria were identified in our analysis of brain tumor samples. In sum, CSI-Microbes offers a new way to identify likely intracellular microbes living within specific cell populations in malignant tumors, markedly extending upon previous studies aimed at inferring microbial abundance from bulk tumor expression data.