Hello,
Would anybody know why in the mouse brain scRNAseq from ALM and VISp cortex regions there appears to be more cells from deeper layers? Also, VISp seems to have more cells in L4 compared to ALM. Is the reason anatomical, technical, or something else?
Thank you!
There is a combination of technical and anatomical reasons for this. Regarding layer 4, ALM is a region that does not have a layer 4–while this region is not explicitly listed in our atlas, it is reasonably close to here. The reason for having more cells sampled from deep layers is likely because we have found a greater diversity of excitatory types in deep vs. superficial layers. By over-representing deep layers we increase our ability to separate these distinct types.
Jeremy,
Thank you very much for the detailed answer!
In that case, what, in your opinion, might be a most comprehensive, perhaps, unbiased towards particular neuron type/class, mouse/human sc/sn RNAseq dataset of the whole brain? So, a comprehensive neuronal cell type atlas, in a way. I know there are fantastic datasets from Linnarsson & team, Tabula Muris, and some other - however, each of them seems to have some limitations.
I’m also aware of a great work you did in the Hodge & al preprint 2019.
What would you suggest may be the best available dataset so far? My purpose is to assemble a sc/snRNAseq dataset with most comprehensive span of brain regions and (transcriptionally) defined neuronal cell classes.
Thank you!
Hi @ainej,
Unfortunately a comprehensive human sc/sn RNAseq dataset of the whole brain does not currently exist as far as I know, although we are generating some data along these lines in collaboration with Sten Linnarsson’s lab, which will not be completed this year. However, in October we will be releasing data and associated cell types from several cortical regions. My field of study is cortex, so I am unsure whether single cell/nucleus human data sets exist from outside cortex.
For mouse, there are more options as you’ve mentioned, and I don’t have enough experience with this data to say which is best. Currently, we have data from a few cortical areas available for download, and data and cell type calls from additional cortical and hippocampal areas will be available in our October release. We also have ~1.5 million cells from multiple mouse brain areas available for download through the BICCN cell registry, although as far as I know they are not clustered. In addition to Tabula Muris and the recent study from the Linnarsson lab, there is at least one large study of developing mouse brain. Finally, here is a slightly older review which describes some older studies of brain cell types.
Hello ainej,
I’m curious how you generate that plot. Did you have a piece of that code?
I’m searching for a layer characterization of the scRNAseq-identity of the neurons
Thank you!
Hi both,
@ainej - there now is a a whole human brain atlas available from the Linnarsson lab, in collaboration with the Allen Institute, if you are still looking for one (here). All the data is available via their GitHub repo as well.
@SamyCN89 - ainej may have specific code, but the information you need to generate these plots are part of the metadata (sample information) available for download. Each cell has some information about layer dissection. Depending on the specific transgenic line (also in the metadata!), individual layers, sets or layers, or the whole cortical sheet were dissected per experiment. In particular, for some of the broader transgenic lines (e.g., pan-neuronal, pan-GABAergic), individual layers were dissected, so you’d be able to characterize the layer specificity per type. This information is also available on the main page of the RNA-seq Data Navigator if you scroll down to the dataset overview.
Best,
Jeremy