De novo spatiotemporal modelling of cell-type signatures in the developmental human heart using graph convolutional neural networks

• DOI: 10.1371/journal.pcbi.1010366
• Publication Date: 2022-08-12T17:47:49Z
• Authors: Sergio Marco Salas; Xiao Yuan; Christer Sylven; Mats Nilsson; Carolina Wählby; Gabriele Partel
• Keywords: Cell type; Single-Cell Analysis; Functional Diversity

With the emergence of high throughput single cell techniques, understanding molecular and cellular diversity mammalian organs have rapidly increased. In order to understand spatial organization this diversity, data is often integrated with create probabilistic maps. However, targeted typing approaches relying on existing achieve incomplete biased maps that could mask true present in a tissue slide. Here we applied de novo technique spatially resolve characterize situ sequencing during human heart development. We obtained made accessible well defined cell-type fetal hearts from 4.5 9 post conception weeks, not by approaches. our analysis, previously unreported within cardiomyocytes epicardial cells identified their characteristic expression signatures, comparing them specific subpopulations found RNA datasets. further characterized differentiation trajectories cells, identifying clear component it. All all, study provides novel for conducting spatial-temporal analyses developmental samples useful resource online exploration development at sub-cellular image resolution.