Can 5' and 3' scRNAseq be processed with the same pipeline?

Question

[ev97]

I have encountered a public dataset where they have generated scRNAseq using the 10x Chromium with both 3’ and 5’ chemistries. Since I am used to work with 3' and I do not know anybody who has worked with 5' to ask them, I wonder if I have to take special attention with the 5' data or if I have to do something different as I usually do.

I have tried to search for information... but I did not find anything about the processing (if there are differences).

I have read that the difference between them is the technique and aim of use.

• In 3’ scRNA-Sequencing, poly(A)+ mRNAs are reverse transcribed using a polydT primer. This polydT sequence is located directly adjacent to the barcodes required for the identification of unique cells and for sequencing.

• In 5’ scRNA-Sequencing, poly(A)+ mRNAs are reverse transcribed using a polydT primer. However, in contrast to 3’ scRNA-Sequencing, the sequencing barcodes are not adjacent to the polydT primer, but they are located at the 5' end of the transcripts. Moreover, I read that the major advantage and use of the 5′ is to combine gene expression profiling with the identification of the full V(D)J fragment (a and b chain of TCR, heavy and light chain of Igs and BCR in B-cells) in the same cells.

... but regarding the processing (if some particular steps need to be taken into account), I didn't find anything.

Does anybody can help me, please?

Any feedback will be well received.

Thanks in advance

Answer 1

[yura.grabovska]

Do you have samples that should in theory have a good degree over overlap in terms of cell types across the two sequencing types?

I would probably recommend an exploratory analysis where you run the standard vignette-type bash and see what the resulting dimensionality reduction looks like and then think about what kind of integration you might need to carry out.

You could look at expression of something like housekeeping genes to see if there's any skew towards a single technology. Depending on the size of the libraries and the total number of cells it might not take to long to have a preliminary look?

From experience I know there are differences between sc/snRNASeq but all the data I looked at was 3'. I would assume that there would be some inherent batch effect between the 3'/5' libraries, but until you visualise it in some way you won't know how much it affects your data.

[ev97]

Thanks very much for your feedback! I will take all those things in consideration once I start the analysis!

Just out of curiosity, do you usually check all those things between sc/snRNAseq too? (overlap of cell types, housekeeping genes...)

[yura.grabovska]

sc/snRNASeq tends to have differences in things like mitochondrial genes etc, which you can account for using tools like SCTransform but in general I just keep it in the back of my mind that there's likely a difference and I look the clustering results to see if there's likely an effect.

[ev97]

Noted! Thanks very much for your help!