human brain data

[antoine4ucsd]

Hello I am very interested in evaluating your package on human brain scRNAseq samples. I tried it with all default (seurat object as input) but it only identifies macrophage and endothelial cells and no microglia, neurons or dendrocyte/astrocytes... are you aware of a human brain atlas we could use as a reference to be more specific?

thank you!

[JiaLiVUMC]

Thanks for your interest.

If you used the pre-defined panglaodb database, microglia is a sub-celltype of macrophage. It is possible that there are macrophages but no microglia identified. But astrocytes or neurons are sub-celltype of Neural cell (first level). This can happen if there are no cluster assigned to Nerual cell in the first level.

Before tried a new reference, maybe you can try these potential solutions firstly:

1. Since you used seurat object as input, scMRMA will use the 'count' matrix as input and then do normalization. You can check the 'count' matrix in seurat object is 'real count'.
2. Maybe you can try to provide clusters by yourself using the 'selfClusters' parameter. There is an example in the last part.
3. If all failed. Maybe you can set up your own reference only includes the celltypes and genes you are interested. But I think the two clusters should be easy to identified and no need for new reference.

Hope it works!

[antoine4ucsd]

thank you for your input! I will give it a try and let you know how it goes.