search

single-cell-genetics / vireo

Demultiplexing pooled scRNA-seq data with or without genotype reference
https://vireoSNP.readthedocs.io
Apache License 2.0
73 stars 27 forks source link

Multiome data demultiplexing #63

Closed zhijunyuu closed 2 years ago

zhijunyuu commented 2 years ago

Hey,

Thanks for the nice tool.

I have some 10x Multiome data, two donors were pooled for one reaction. In this case, I use cellSNP-lite and vireo to demultiplex scRNA and scATAC data independently. Most of the cells are assigned consistently within these two datasets, but some of them are not. Is there any way to to the demultiplex together via combining scRNA and scATAC data regarding they come from the same cell? Thanks in advance!

Here is the eg: 1) result from scATAC data: Var1 Freq donor0 1877 donor1 2416 doublet 370 unassigned 902

2) result from scRNA data: Var1 Freq donor0 1844 donor1 2707 doublet 322 unassigned 692

Best, Zhijun

huangyh09 commented 2 years ago

Thanks for sharing. It's a good point to combine the two modules, particularly for reducing the unassigned cells.

However, Vireo doesn't have a function to do so. Here are two potential options:

  1. Append the SNPs between the two omics, and ignore that they may come from the same SNP. It may have the computational convenience to simply append the sparse matrices between to modules.
  2. Sum the matrices between the two omics, if you keep the same set of SNPs. One limitation could be that UMI is used in scRNA but not in scATAC module.

Yuanhua

zhijunyuu commented 2 years ago

Hi Yuanhua,

Thanks for the reply. I tried append the matrix (output from cellsnp-lite) from the two omics and run vireo. It works! Here is the result, I think it is already much better than single dataset regarding the reduced number of unassigned cells:

Var1 Freq donor0 2929 donor1 2181 doublet 416 unassigned 39

Best, Zhijun

huangyh09 commented 2 years ago

Nice and thanks for sharing. I'll close this issue here but feel free to re-open it.

Yuanhua

DHelix commented 7 months ago

Hi Yuanhua,

Thanks for the reply. I tried append the matrix (output from cellsnp-lite) from the two omics and run vireo. It works! Here is the result, I think it is already much better than single dataset regarding the reduced number of unassigned cells:

Var1 Freq donor0 2929 donor1 2181 doublet 416 unassigned 39

Best, Zhijun

Hi @zhijunyuu, I've been working on merging scRNA and scATAC data from scMultiome data for demultiplexing. Could you please explain how you appended the cellsnp-lite output matrices? Thanks a lot!