Cell origin deconvolution pipeline for scRNA-seq data
Thatchayut Unjitwattana, MS, Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI; Qianhui Huang, MS, Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI; Youqi Yang, MS, Department of Biostatistics, University of Michigan, Ann Arbor, MI; Mengtian Zhou, MS, Department of Statistics, University of Michigan, Ann Arbor, MI; Lana X. Garmire, PhD, Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI
Poster # 73
Single-cell RNA sequencing (scRNA-seq) is crucial for understanding cellular mechanisms at single-cell resolution. However, for tissues with complex origins (such as placenta) or prone to contaminations (e.g. blood in tumor samples), rigorous preprocessing needs to be done before the downstream functional analysis. We propose a new pipeline to decipher the origin of cells which utilizes inferred genotype (maternal vs. fetal) information from scRNA-seq data to deconvolute barcoded cells into different origins and separate blood cells from resident tissues. Using the simulated data, we show that the proposed pipeline can deconvolute tissue and blood residents with high efficacy. We further demonstrate the utilities of this pipeline by applying it to adenocarcinoma (PDAC) and placenta tissues prior to the downstream analysis. In PDAC tissues, this pipeline deconvolutes cells into tumor and adjacent normal origins, as well as blood- and tissue-resident cells. This provided insights into differences in activities and signaling pathways of immune cells with respect to their microenvironment. In placenta tissues, this pipeline successfully deconvolutes cells into maternal and fetal origins together with blood and tissue residents. It reveals cellular and immunological differences between placenta tissues of maternal and fetal origins without the contamination from both maternal and fetal blood cells, which could not be accomplished without this pipeline. Taken together, we propose the pipeline to systematically analyze tissues based on different origins and blood-tissue residents which effectively reduces artifacts including mixed cell types that are not of interest.