A deep learning approach enables fast and robust prediction of hematopoietic stem cell lineage choice in time-lapse imaging three generations before conventional marker onset. Differentiation alters molecular properties of stem and progenitor cells, leading to changes in their shape and movement characteristics. We present a deep neural network that prospectively predicts lineage choice in differentiating primary hematopoietic progenitors using image patches from brightfield microscopy and cellular movement. Surprisingly, lineage choice can be detected up to three generations before conventional molecular markers are observable. Our approach allows identification of cells with differentially expressed lineage-specifying genes without molecular labeling.

Haematopoietic stem cells (HSCs) are the founding cells of the adult haematopoietic system, born during ontogeny from a specialized subset of endothelium, the haemogenic endothelium (HE) via an endothelial-to-haematopoietic transition (EHT). Although recently imaged in real time, the underlying mechanism of EHT is still poorly understood. We have generated a Runx1 +23 enhancer-reporter transgenic mouse (23GFP) for the prospective isolation of HE throughout embryonic development. Here we perform functional analysis of over 1,800 and transcriptional analysis of 268 single 23GFP⁺ HE cells to explore the onset of EHT at the single-cell level. We show that initiation of the haematopoietic programme occurs in cells still embedded in the endothelial layer, and is accompanied by a previously unrecognized early loss of endothelial potential before HSCs emerge. Our data therefore provide important insights on the timeline of early haematopoietic commitment.