Compare bulk RNA-seq with single cell RNA-seq

[icanwinwyz]

Hi everyone,
I recently used 10X chromium techniques to do the single cell RNA-seq for my sample. I totally yielded about 3000 cells. I would like to compare the single cell transcriptome profile with the bulk RNA-seq using the same sample to check the consistency (correlation). However, 10X used the 3' sequencing strategy and produced UMI counts but bulk RNA-seq sequenced the whole transcript and produced RPKM. I would like to ask does it make sense to check their consistency with different sequencing strategy/units? If so, what is the best way to do this?

Many thanks in advance!

[nucacidhunter]

I have heard presenters stating this to convince that their single cell data is reliable. I do not see any reason for this justification because of following:

1- Bulk RNA-seq data resulting from the same sample using polyA capture and rRNA depletion are not highly correlated

2- Bulk RNA-seq data resulting from the same sample using two different kits are even less correlated

3- Assuming 10pg RNA/cell your combined total RNA input would be 30 ng. Most bulk RNA-seq kits require minimum 100 ng input so the results will not be comparable.

[cmbetts]

3- Assuming 10pg RNA/cell your combined total RNA input would be 30 ng. Most bulk RNA-seq kits require minimum 100 ng input so the results will not be comparable.

Generally once you're >30 cells the average expression starts to converge on bulk sequencing (Prepared by the same protocol). The issue here is that you can't recreate the 10X protocol on your bulk sample. In my experience with a different 3' scRNA method, you'd be lucky to get a correlation >0.7 compared to bulk conventional RNA-Seq.

[nucacidhunter]

Generally once you're >30 cells the average expression starts to converge on bulk sequencing (Prepared by the same protocol).

I wonder if there is any reference for this.

[skannan4]

I found this paper that suggests that at >10 cells expression converges to bulk (r = 0.92). But just to follow up, do you all suggest that there is no need to run a bulk sample as QC when running single cell RNA-seq? Obviously we would like to drop the sequencing costs if possible and not needing to sequence the bulk sample would be a relief.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4191988/