Hi there, I think I have a very basic question. I am trying to use the Mouse whole cortex and hippocampus 10X dataset to do compare expression of cholinergic receptor genes acorss interneurons classes and excitatory cells in posterior parietal cortex (VISa and VISrl). I’ve previously done this for VISp using the V1 & ALM - SMART-seq dataset, but with the whole cortex dataset I can’t figure out how to get started, specifically, how to isolate and download just the data I need (specific genes in a specific region for specific cell types). Thanks!
1 Like
Hi @CelineCammarata,
We get asked these types of questions quite frequently. In short, there is no way to download only a subset of the data. You have to download the entire data set and subset it yourself (both for genes, and for brain regions). There are several posts about this data set, but I’d recommend starting here: Gene expression matrix.cvs is too large to load it - #17 by jeremyinseattle.
An alternative approach would be to use the Transcriptomics Explorer to view the data on-line. First, the Sampling Strategy visualization could show you which clusters contain cells from VISa and VISrl (or at least which cells are from VIS but not VISp… there are no regions defined as VISa or VISrl in the online visualization).
Next you could view your genes of interest, and see which clusters they were expressed in. In this example, I added a few random genes, but you could instead add the ones you care about.
Hopefully this is enough to get you started.
Best,
Jeremy
Hi @jeremyinseattle
I’m going through a similar conundrum with the ABC atlas where I’m interested in identifying the proportion of GeneX- and GeneY-expressing cells within a given cluster/dissection region.
It’s great that we can filter by dissection area/cluster and then further filter by GeneX-expressing cells within a given area/cluster. Using this methodology one can estimate the proportion of GeneX-expressing cells of a given cluster/dissection region.
However, it looks like you can’t use more than one gene filter at a time.
If I want to calculate the proportion of GeneX+ and GeneY+ expressing cells within a given cluster, would I need to download the raw data?
One method would be to select and download the cell names of the entire cluster, then GeneX+ cells, and finally GeneY+ cells after filtering by cluster and then compare cell identities. However, we’re limited to downloading only the first 100 cells so this doesn’t sound feasible.
Greatly appreciate any guidance, Thanks!
Jesse
Hi @jkniehaus,
I think your best (and possibly only) option as of right now if you’d like to do multi-gene filtering in the way you suggest is to download the data. The best way to do this is to follow the Jupyter notebooks available here: Allen Brain Cell Atlas - Data Access — Allen Brain Cell Atlas - Data Access.
These should be self-explanatory, but if they are not, please let us know how we can improve them to make the data more accessible.
Best,
Jeremy
Hi Jeremy
I think we’re making progress! Now my next question is, what is a good way to identify interneurons from a particular layer? For example, if I want to find SST-expressing cells in only layer 2/3?
Thanks,
Celine
Hi Celine,
You can do this manually in the ABC Atlas for the whole mouse brain data set by adjusting the filters (like this example). Otherwise, you’d probably need to download the MERFISH and scRNA-seq data and do some coding. If and when imputed gene expression data for MERFISH becomes available, this will be a bit easier.
For the cortex + hippocampus data set, some of the data were specifically dissected from different cortical layers, so you could see which cells come from layers 2-3:
Given how varying the dissections are and the fraction of cells without dissected layers, this also may not be straightforward, but by eye there do appear to be a few types that show enrichment in superficial layers.
A final option is to read the papers, which I’d expect this was discussed 
.
Best,
Jeremy