Assigning Allen Transcriptomic Taxonomy to External Dataset


How can Allen transcriptomic identities be assigned to external datasets. For example assigning nuclei from a query dataset as Sst Myh8 vs Sst Chodl?

I have tried using the Allen data as a reference and then mapping my cells onto that reference using Seurat MapQuery. The issue is that this requires clustering the Allen data with Seurat to create the reference and this does not produce the exact same clusters/cell types as the found in the papers. For example, I will get a cluster composed of multiple Sst subtypes (based on the Allen metadata) so when a query cell maps to that cluster I am still not sure what subtype it is.

Is there a better way to do this? Can Scrattch.hitcat do this and if so how?

Thank you!


If you’d like to use Seurat, I would recommend using the function TransferData, which takes as input the data AND cell type assignments from a reference data set and transfers (maps) those labels into your query data set (see the examples section at the bottom of the page). Also, if you happen to be interested in adult primary motor cortex, both mouse and human data sets from this region are included in Azimuth, which is a user friendly web application from the Satija lab that has Seurat as its engine. We have tested several other mapping algorithms and hope to have something to show in this space by the end of the year.

Azimuth is great! Unfortunately I am interested in mouse primary visual cortex. Is there any plans to add the Allen V1 data to Azimuth anytime soon or at all? In the mean time I will try using TransferData.

Glad to hear. Details of specifically what we will be releasing and when for mapping tools are still under discussion but we’ll post about it on the Community Forum and/or the main page when something is released.

You also asked about mapping in scrattch.hicat and it seems the answer is yes because there is a section about it in the tutorial.