Using single cell technologies, we profiled the transcriptome of 486,134 cells and nuclei from six anatomical cardiac regions. Given the complexity of these data, we integrated them using state-of-the-art machine learning methods. Using this strategy, we identified 11 major cardiac cell types and 62 different cell states.
We collected healthy hearts from 14 adult donors, seven male and seven female, within the age range of 40 - 75 years. The hearts were collected in North America and the United Kingdom.
Using a variational autoencoder, we integrated the single nuclei produced in two different institutes, the total single cell and the CD45+enriched single cell into a unified manifold by learning the latent representation of each data source.
Single nucleus and single cell sequencing capture different cell types. Single nuclei captured most of the cardiomyocytes, fibroblasts and adipocytes; whereas single cell/CD45+enriched single cell capture most of the vascular and immune cells.
Unsupervised clustering of the cardiac cells reveal stark regional differences in specific cell types. The most dramatic one being the transcriptional difference between atrial and ventricular cardiomyocytes.
We also observe transcriptional differences by regions for fibroblasts and vascular cells.
Combined single cell and single nuclei RNA-Seq data of 485K cardiac cells with annotations