SEA-AD Multiome problems

How To Allen Brain Cell (ABC) Atlas
, ,
1

Hi Guys! I was checking the Seattle Alzheimer’s Disease Brain Cell Atlas GitHub page and down at the data location part, I found this:

“Briefly, the processed single nucleus RNAseq, ATACseq, and Multiome as well as MERFISH datasets are available from the AWS Open Data Registry.”

They say I can find Multiome data in this aws link/bucket, but I haven’t. Now, I am trying to analyse DLPFC Multiome data, but I’m taking GEX samples and ATAC samples manually, which probably is not the best way possible, because I’m not ensuring the perfect matching. Can anyone help me on this topic?

Thanks!

2

I believe that you are doing it the right way; however, I’ve also reached out to folks more closely involved in the data analysis who will reply with a potentially more efficient method when they are back in the office.

1 Like
3

Brilliant! Thank you, Jeremy.
I was testing my pipeline with scDoRI (Welcome to scDoRI: Single-cell Deep Multi-Omic Regulatory Inference — scDoRI documentation), but I was getting extremely strange UMAPs and GRNs using this tool. That’s why I considered two possibilities:

  1. I might be incorrectly matching the multiome data (RNA + ATAC), or
  2. (More likely) I might be making a mistake in the annotation step.

Regarding this second possibility, I was actually going to ask if the data analysis team could share the link to the reference dataset they used for annotating the multiome data.

4

The short version is that the MTG data from the AWS bucket was originally annotated by aligned to these data from neurotypical younger donors form here (Human MTG 10x SEA-AD - brain-map.org; also a subset of data in the AWS bucket), with non-neuronal cell types expanded via clustering. DLPFC data was then integrated with the MTG data and assigned MTG types. If I’m remembering correctly, this was all done initially using the RNA-seq data only (both single-ome and multiome), and later cell types from the single-ome ATAC-seq cells were assigned through integration with the multiome data. In all cases, multiome cells should have the same annotations for the corresponding RNA-seq and ATAC-seq data, since it is the same cell.

1 Like