Would you please help with these questions? Is this a reference cell types for human middle temporal gyrus? https://portal.brain-map.org/atlases-and-data/rnaseq/human-mtg-10x_sea-ad
I don’t know why in subclass_label column, beside cell types like astrocytes, endothelial cells, we also have Sst, Lamp5, etc. Seem they are marker genes. Here are list of all unique values in subclass_label which I think cell type column:
Pax6, L5/6 NP, L5 IT, L6 CT, L4 IT, Astrocyte, L2/3 IT, Vip, Sst Chodl, L6 IT Car3, L6 IT, Sncg, Pvalb, Oligodendrocyte, Lamp5 Lhx6, Sst, VLMC, Lamp5, Microglia-PVM, OPC, L6b, Endothelial , L5 ET, Chandelier, max
Thank you so much!
Yes, this is a reference for cell types in adult human MTG, and is the starting point for the cell type taxonomies used in SEA-AD. We use different conventions for naming different types of cells in cortex:
For GABAergic interneurons, cell types are organized based on expression of gene historically used to group them, since they do not as nicely align to specific cortical layers as do glutamatergic interneurons. So in this case the genes “Lamp5” actually represent cell type designations.
We use the same cell type names across species. I’m not sure if we are consistent in upper vs. lower case in all studies in all species, but assume that “PVALB ###” in one study and “Pvalb ####” in another studies are both the same kind of cells (although matched or mismatched ###'s after the “Pvalb” does not necessarily mean or not mean those are matched subpopulations of Pvalb interneurons).
PVM stands for “brain PeriVascular Macrophages”, and so the Microglia-PVM group essentially includes all immune cells.
In our SEA-AD MERFISH data, both cell types are relatively common, so I’m not sure why you aren’t seeing them. It could be the difference between cell-based and spot-based spatial transcriptomics methods, if the overall abundance (and therefore gene expression levels) of these types are low.
Hi @jeremyinseattle. The dataset in the paper used visium data and annotated cell type by a tool name cell2location, and I use this reference: Human MTG 10x SEA-AD - brain-map.org, not MERFISH. Or it is hard to annotate cell type that scatter in tissue like astrocyte and microglia using seurat transfer method? The seurat vignette annotated visium data but for cell types that in layer but not scatter. Maybe that is why the author used cell2location. I really appreciate your help!
With spot-based methods (like Visium), depending on the spot size, you have a cell deconvolution problem. Others could speak to this better, but typically with these types of problems the more abundance a cell, the more accurate you can assess. For example if microglia represent (and I’m making this number up) 1% of the cells in a spot, there is very little signal overall to find. I expect your intuition about scattered cells being harder to annotate is accurate, but I’m not sure. Best of luck in your analysis.
