SSMD: A semi-supervised approach for a robust cell type identification and deconvolution of mouse transcriptomics data

ABSTRACT Deconvolution of mouse transcriptomic data is challenged by the fact that mouse models carry various genetic and physiological perturbations, making it questionable to assume fixed cell types and cell type marker genes for different dataset scenarios. We developed a S emi- S upervised M ouse data D econvolution (SSMD) method to study the mouse tissue microenvironment (TME). SSMD is featured by (i) a novel non-parametric method to discover data set specific cell type signature genes; (ii) a community detection approach for fixing cell types and their marker genes; (iii) a constrained matrix decomposition method to solve cell type relative proportions that is robust to diverse experimental platforms. In summary, SSMD addressed several key challenges in the deconvolution of mouse tissue data, including: (1) varied cell types and marker genes caused by highly divergent genotypic and phenotypic conditions of mouse experiment, (2) diverse experimental platforms of mouse transcriptomics data, (3) small sample size and limited training data source, and (4) capable to estimate the proportion of 35 cell types in blood, inflammatory, central nervous or hematopoietic systems. In silico and experimental validation of SSMD demonstrated its high sensitivity and accuracy in identifying (sub) cell types and predicting cell proportions comparing to state-of-the-arts methods. A user-friendly R package and a web server of SSMD are released via https://github.com/xiaoyulu95/SSMD . Key points We provide a novel tissue deconvolution method, namely SSMD, which is specifically designed for mouse data to handle the variations caused by different mouse strain, genetic and phenotypic background, and experimental platforms. SSMD is capable to detect data set and tissue microenvironment specific cell markers for more than 30 cell types in mouse blood, inflammatory tissue, cancer, and central nervous system. SSMD achieve much improved performance in estimating relative proportion of the cell types compared with state-of-the-art methods. The semi-supervised setting enables the application of SSMD on transcriptomics, DNA methylation and ATAC-seq data. A user friendly R package and a R shiny of SSMD based webserver are also developed.