Hi,
I was successfully able to use MapMyCells, which is a tremendous tool for the community.
The output shows cell class and subclass labels for each mapped cell. However, I am more interested in predicting regions where the cell may be matching to.
Is it possible to include this in the MapMyCells output?
Since the region labels for each cell are available, it seems like adding additional columns much like the (sub)class columns could be added for predicting the brain region.
Looking forward to reading your thoughts.
Best,
Koen
Thank you for your interest in MapMyCells and for your suggestion on how we can improve it to better serve you. Mapping data to regions is not currently supported, however, we will consider adding this capability in a future release. In the meantime, please let me know if you have any other questions.
Best,
Elysha Fiabane
Product Manager, MapMyCells
Hi @Elysha-AIBS ,
Thank you for your response.If not supported using MapMyCells, how would you suggest to approach this using the general Allen Brain Atlas data and related tools?
I suppose downloading the > 4 million cells locally and leveraging their corresponding annotation using a transfer learning approach would be overkill: can you think of a more permissive strategy, based on available tools from the Allen Brain Atlas?
Hi,
Have you find the solution? I am in the same boat.
thank you
While this is not a direct answer to the challenge of mapping to brain regions, it is worth noting that many cell type show anatomical localization. (Here is a recent example I’ve found that is almost entirely in layer 4 of somatosensory cortex). After using MapMyCells for mapping, the ABC Atlas would allow visual inspection of potential anatomic specificity of mapped cell types. If you are comfortable with python, the ABC Atlas Jupyter notebooks may allow a more systematic way to identify regions for mapped cell types. Finally, this file adapted from the original manuscript has columns summarizing anatomic specificity of each cell type.
To reiterate: this is not a direct mapping of user data to mouse anatomic structures, but is the approach I would take given existing tools, and will be useful for a subset of brain cell types. A similar approach could be taken mapping to whole human brain.