Human brain MTG single cell data mapped to mouse brain single cell RNAseq?

Is anybody aware of a publication where human brain MTG snRNAseq dataset is projected onto mouse brain scRNAseq dataset entirely as opposed to projecting just specific cell classes (excitatory or inhibitory neurons only, etc)?

This is an interesting analysis question worthy of a whole publication. For some experiments and analysis that have covered this topic, refer to publication references in the Cell Taxonomy section of the Allen Institute for Brain Science website, here: This recent publication by Bakken, Hodge, at al may be interesting to you.

Dear Amy,
Thank you very much for your response! I was reading Hodge et al 2018 preprint in detail.
Would it be fair to extrapolate that by enriching for the specific neuronal class (excitatory for example) in human MTG the dataset would contain most (if not all?) excitatory neurons in human brain across brain regions?
My ultimate question for this pre-print is the following: when mapping the human MTG dataset to mouse scRNAseq, say, specifically for excitatory neurons, how confident we are in the exhaustiveness of each of the dataset (approximately what fraction of all possible human exc. neuron cell types is represented in the human data, and similarly, in the mouse data), and hence in the resulting equivalence of the cell types between the species?

Thank you very much for your help and for an interesting discussion!

Hi @ainej,
These are great questions and our ongoing lines of research are also trying to address them! IMHO, I doubt very much that we have ‘all’ excitatory cell classes represented in human brain based on what we see from the MTG data. I would expect most high level classes to be captured, but we can’t know what we are missing without actually being able to compare with other areas - we are actively gathering more data to help address this. For the most up-to-date access to raw data, consider visiting a collaborative repository of data being generated by the NIH BRAIN Initiative consortium, BICCN (we are part of this). NeMO is hosting RNASeq data from many labs, here: But the data here doesn’t include interpretative analysis, which is in progress.

So: In considering the Hodge et al paper, your question is THE open question for us, also. Drs. Hodge & Bakken, along with many other scientists and data analysts at the Allen Institute, are actively working on this very question and I would you encourage you to seek their newest work out at the Society for Neuroscience conference this year, if you are attending. There, we will have the scientists’ and their posters, presentations and workshops related to botth human and mouse transcriptomics; learn more from our events page.

1 Like

Hi @ainej,
To further address this question, I agree with @amyb that we haven’t yet sampled all of the excitatory types in human at the same resolution as in mouse. As a notable example, the cell type defined as “pyramidal tract” in our recent preprint is clearly undersampled. This type is significantly more rare in human than mouse and we have no published way to enrich for this type yet (but our mouse team has a promising enhancer in the works–see Figure 3 here). We have some evidence that these cells will subdivide into more types in human with additional sampling. Please visit our site again after our October release, when we will share our cell types defined using multiple cortical areas in both mouse and human.