Comparative annotation of porcine bone marrow cell types using scRNAseq

Abstract

Abstract Pigs are important as a biomedical model and for worldwide meat production. Therefore, a better understanding of porcine hematopoiesis and immune system development could have a wide impact on human disease research and food production. The bone marrow represents a unique environment that is the major site of differentiation of stem cells and progenitors into mature blood cells. We performed scRNA-seq on bone marrow from healthy adult pigs and sequenced a total of 3,576 cells. After QC, 2,707 cells and 16,432 genes were used for downstream analysis. The porcine bone marrow cells were grouped into 15 distinct clusters and these relationships were mapped using non-linear dimensional reduction. To find markers that defined each cluster, we found a total of 3,335 differentially expressed genes using the non-parameteric Wilcoxon rank sum test. Further, pairwise differential expression testing and random forest models were used to evaluate the similarity between clusters and further define the unique features of each cluster. Using these defined markers, we were able to annotate the cell types for each cluster. The clusters represented a diverse set of cell types including myeloid and lymphoid cells, myeloid and lymphoid cell progenitors, and platelet and red blood cell progenitors. We validated these annotations using publicly available datasets including the CITE-seq database of human bone marrow mononuclear cells (Stuart et al., 2019). We also compared the expression profiles between similarly annotated cell types between human and pig. Our novel annotation of porcine bone marrow cells will be a resource for the study of porcine hematopoiesis and immune system development and help inform human translational biomedical research.