Does it make sense to use scRNAseq for this experiment?

Question

[Mark]

I’ve never done scRNAseq so I don’t know about if there are any pitfalls that can derail or prevent this analysis from being successful for this specific application but our lab does a lot of metabolic engineering in rare microbes. We transform genes from biosynthetic pathways into the organisms and try to get the organisms to express valuable biosynthetic compounds that we then extract and purify. One thing is that after the transformation of certain genes, we get unexpected results or lack of expression that we expect and we have trouble finding out why. Our lab is trying to incorporate more bioinformatics approaches to get a more holistic view of what’s happening in these organisms and one of the approaches we’re considering is scRNAseq to see how expression of our genes of interest and all the cells other genes are affected by our transformations. I’m wondering if this is a good approach. From what I can tell, a lot of use cases for for scRNAseq are biased towards human and mouse research in order to differentiate expression profiles in different tissues. This isn’t as relevant for our purposes since we’re working with microbes and I’m thinking the main difference in “cell types” in our sample might be the cell cycle of the individual cells. Additionally, our cells have very little external resources out there (e.g. lack of good gene models, RNAseq data, biological pathway information, etc), so if there’s steps that require good reference data to really perform a good differential gene expression analysis using scRNAseq it would be good to know that ahead to time to know if this approach should be avoided. If scRNAseq doesn’t seem like the right tool for the job, can anyone tell me the reasons why and if possible what approach they would recommend instead? Thank you!

Answer 1

[i.sudbery]

Single cell RNA-seq from bacteria is quite unusal, I'm not sure which kits or protocols support it, although I believe it has been done at least once before.

The key question to ask yourself is what do you gain from doing scRNAseq that you wouldn't get from doing bulk RNAseq? Bulk RNAseq is many times easier (you just isolate RNA and send it off to a company) and cheaper (<$200 a sample vs thousands) and gives you more data.

For example, the lack of good genome/annotation is not necessarily a problem with bulk RNA seq, because you can annotate you own gene models using the RNA-seq data. If you don't have a genome, you can even do de-novo assembly of the RNA-seq to generate a transcriptome without having to have a genome. This would be much harder from scRNAseq.

In general scRNAseq is useful when you are interested in some aspect of the heterogeneity of the cells in your sample - if you are interested in what sub-populations are present, or how much cells differ from each other, or you are only interested in a rare sub-population that can't be isolated any other way. For most other experiments, I'd almost always recommend bulk RNAesq.

One thing to be careful about is the ribosomal RNA depletion. This is done using oligo pulldowns, and the oligos in the standard kits are designed to target mammalian rRNA. Bacterial rRNA depletion kits exist, but you need to make sure you ask your service provider to use them. If your bug is very different from the bugs used in the kits (I'm pretty sure it works for e.coli, Strep, Staph and Clostridium), then a custom set of pull down oligos might be needed. Alternatives you could just do a lot of sequencing and throw the majority (which will be rRNA) away. A lot here is relative. You'd probably still need much less sequencing than a standard human sample, and the major cost is going to be library prep, not the sequencing.

[Mark]

Thank you that's a very good answer