Single Cell RNA-seq and validation

Question

[Genosa]

Hello all,

I would like to ask for expert advice here about single Cell RNA-seq to study differential gene expressions in treated vs. untreated groups especially about sample size. According to the nature rev genetics paper by Shapiro 2013, he recommends a minimal of 50 cells are needed to reduce the error for quantitative transcriptomic analysis. Then I come across other papers / methods such as EdgeR, Scotty which seems to suggest smaller size of 12. Of course, provided cost and sample availability is NOT a concern I would just go for '50' but I'm not sure if it is safe at all to just sequence 12 cells and hope for statistically meaningful results.

Next is about validation - I've seen methods such as single cell QPCR, digital PCR, single cell nano string. Which method would be suitable for validation, and again how many single cells would you pick to validate? Thank you very much in advance for your helpful advice.

[Sam]

I have never do single cell sequencing before but that 12 sample suggestions of EdgeR does sounds like it is for normal RNASeq instead of single cell RNASeq. So maybe you should becareful about whether if that suggestion was based on single cell or not?

[Genosa]

Thank Hi Sam,

Thank you very much for your help. That EdgeR software was not used on single cells, and so far no single cell RNA seq literature quotes this. My problem is because my samples are very limited, at most is 50 (if lucky), I don't know if it's better to do just triplicates of bulk population vs. single cell sequencing of these 50 cells.

Answer 1

[LabCoatNomad]

what sequencing machine platform and machine do you plan on using for single cell?

if you are doing illumnia NextSeq or NovaSeq keep in mind that under loading a lane can be just as bad as over loading a lane. 50 cells is far short. You will either need a lot of samples at 50 cells each OR pool with someone else you know running bigger samples..

because of the very small cell size i have to also ask, how homogeneous are these cells? purified? sorted?

It depends on the biological question being asked of course but the cost for bulkseq is much cheaper, and if your question can be answered by it, you might benefit from going that route for so few cells. SOME of the costs of single cell will not be reduced very much if you are running 10,000 cells or 50 cells... but if you have a lot of 50 cell samples might still be worth it...

As for the 2013 bulk seq recommendation, bulk seq has come a long way in 7 years, generally different technologies have different minimum requirements for RNA to process, they vary by kit & tech, but also note that different cell types have varying amounts of RNA , so also know that one persons protocol might not apply to you depending on cell type and species.