Abstract
Here, we present Spatial Genomic Analysis (SGA), a quantitative single-cell transcriptional profiling method that takes advantage of single-molecule imaging of individual transcripts for up to a hundred genes. SGA relies on a machine learning-based image analysis pipeline that performs cell segmentation and transcript counting in a robust way. SGA is suitable for various in situ applications and was originally developed to address heterogeneity in the neural crest, which is a transient embryonic stem cell population important for formation of various vertebrate body structures. After being specified as multipotent neural crest stem cells in the dorsal neural tube, they go through an epithelial to mesenchymal transition in order to migrate to different destinations around the body, and gradually turn from stem cells to progenitors prior to final commitment. The molecular details of this process remain largely unknown, and upon their emergence, the neural crest cells have been considered as a single homogeneous population. Technical limitations have restricted the possibility to parse the neural crest cell pool into subgroups according to multiplex gene expression properties. By using SGA, we were able to identify subgroups inside the neural crest niche in the dorsal neural tube. The high sensitivity of the method allows detection of low expression levels and we were able to determine factors not previously shown to be present in neural crest stem cells, such as pluripotency or lineage markers. Finally, SGA analysis also provides prediction of gene relationships within individual cells, and thus has broad utility for powerful transcriptome analyses in original biological contexts.