Single-End to reduce cost for Differential Expression Analysis ?

Question

[Evan]

Hi, I would like to know if Single-End for transcriptome sequencing with approximately 25 milions reads per sample sequencing is sufficient for DEG analysis and will not introduce bias ?

I found good information here about the number of reads but I don't know if it's for paired-end or single-end data : https://emea.support.illumina.com/bulletins/2017/04/considerations-for-rna-seq-read-length-and-coverage-.html

Thank you in advance for your answers :)

[Shred]

Technologies like 3'mRNAseq (see https://www.lexogen.com/quantseq-3mrna-sequencing/ for example) give you less sequencing reads (1 read for each mRNA molecule) and work well with model organisms.

[seidel]

sufficient for DEG analysis and will not introduce bias?

It depends on your goals and bottlenecks. What is your organism? What is your goal? If the bottleneck is knowing in general what genes are changing in response to a condition, 25 million SE reads is sufficient for most common research organisms (including humans) to get a gene list to chase down. However, if your goal is to investigate transcript isoforms, or if your organism has much repetitive sequence, gene duplications, homologs, etc. you'll be less effective at distinguishing these things than with paired-end sequencing. So initial experiments to generate hypothesis or foraging through lots of gene expression space (SE usually sufficient), vs. more careful analysis of known genes in a known system will have different considerations. And then funding/how many experiments you can do. PE usually doubles the sequencing cost, which for many people means they do fewer experiments. So goals and bottlenecks. What are yours?

[Evan]

Thank you very much for your answer. We're currently working on human model and this DEG analysis will be oriented on CAR-T cells. As far I know after discuss with the biologists which made the experiments, they would like to get essentially a list of up and down regulated genes between their different conditions (5 at total, triplicates for each). So that's the reasons why Single-End seems to be adequate for this research purpose and they wouln't like to make compromise on their experiments. The fact that 25 millions SE reads is sufficient reassured me because that was the shadow area for me.

I will give an update here once the sequencing will be done :)

Answer 1

[Zhilong Jia]

The impact of read length on quantification of differentially expressed genes and splice junction detection saying "A researcher could save substantial resources by using 50 bp single-end reads for differential expression analysis instead of using longer reads. "

[Evan]

Thank you !

Answer 2

[colindaven]

We as a core unit have moved from 1x76bp SE read profiling to 2x38bp (NextSeq550).

That is, we use the same kits and flowcell to my knowledge, but divide up the chemicals to use the more advantageous paired end reads. Works well because the distance between the reads is "known" so the aligner can place the read pairs on longer genomic regions considering this information.

There is a paper which exhaustively compared this and gave us the idea (of course, I can't find this right now).

We're currently comparing whether this works in metagenomics alignment as well. I have my doubts!