Skip to contents

Data browser

Source: vignettes/articles/data_browser.Rmd
data_browser.Rmd

Select a dataset and a gene to view the spatial expression of that gene beside the cell type distribution. The tables below list the gene’s ESV score for the whole tissue and for each cell type, and its PoPS rank for each trait.

Notes

  • Count is the raw UMI count for that gene, exactly as stored in the dataset — no normalization or transformation is applied to the numbers you see.
  • Visium and CosMx datasets are shown at different resolutions. The Visium sets have a few thousand spots each and are shown per spot. The CosMx sets hold between 181,000 and 1.55 million cells — far more than a plot this size can resolve — so their cells are aggregated onto a spatial grid of roughly 20,000 bins, and each bin reports the mean raw count of the cells falling in it. The panel caption states which resolution is in use.
  • Cell type proportion: For each Visium dataset, cell type proportions were estimated with RCTD, using the CosMx dataset from the corresponding tissue as the reference (e.g., Visium human DLPFC paired with CosMx human frontal cortex). For the binned CosMx datasets, each bin shows the mean proportion across the cells it contains.
  • PoPS rank is the gene’s position when all 17,033 genes in the PoPS table are ordered by their score for that trait, so rank 1 is the top-scoring gene. Ranking is over the whole PoPS table, not just the genes present in the selected dataset. Roughly half the genes score exactly zero for a typical trait; they tie, and all of them take the first rank of that tied block — around 8,400 — which should be read as “no signal” rather than “mid-table”.