Downloading ISH or Expression data

I want to estimate neuron number, neuron density or local neuron density gradient in a specific brain region using a marker gene expression. For example, for Pvalb in Gene Search I get ISH or Expression images like this:
Is this data available in image format only? Are there high-resolution grids which can be used with data masks for a brain region?


I’d like to know that too!

I wanted to do something similar in the past, and in the end I resolved with downloading and processing the images to extract cell locations and do the analysis on this. Unfortunately however the process was very slow and error prone, so it would be awesome to have a gridded high resolution gene expression database.

1 Like


Yes, the level of expression signal in grid format is readily available. See for information on how to download grid data for each image series.

The expression has also been summarized by structure, as described here:

Best regards,

Hi Wayne,

Instructions in the first link lead to the expression grids with 200 um sampling. According to Erö et al. (2018), we have to extract the expression pixels ourselves if we need higher resolution. That’s why I asked the question, actually. I can get a hi-res region mask and combine it with the slice images with ISH/Expression energy data but then I have to count expression spots myself. I thought to avoid that if 10-25 um expression grids were available. Does it make sense?

Thanks for looking into this!

Hello Alex

The 200 um value is a function of the sectioning scheme used for the Allen adult mouse brain atlas. See the documentation for details. Each gene is only assayed on sections that are spaced 200 um from each other, so a finer resolution grid is not possible.

Thank you, Wayne! I see, it is clear now.

If you prefer using R, @lucasg’s R package cocoframer includes some functions for retrieving voxelized data as well, in addition to it’s primary use as a tool for viewing the Allen Mouse CCF.