Read length and sequencing depth for RNA-seq (for differential gene expression analysis) of bacterial RNA

Question

[abhilashtripathi10]

Hello Everyone

My group in my lab wants to elucidate differential gene expression during bacterial biofilm growth on some material surfaces. For the same, I want to know what should be the minimum read length, total no. reads, and whether I should go for PE or SE RNA sequencing. I went through the literature but I find some papers say 150bp PE (40million reads) while others say 50bp SE (20million reads) should suffice. We have a limited budget and any insights or help from anyone in this community in this regard would be greatly appreciated. Thank you.

[GenoMax]

https://genohub.com/recommended-sequencing-coverage-by-application/

Since you are working with bacteria, you can stay on the low end of estimates there. Be sure to include biological replicates (3 min more better) to ensure your results have statistical support.

[abhilashtripathi10]

Thank you for the suggestion. I am trying to be on the low end with more biological replicates as you suggested.

Answer 1

[ATpoint]

In general, since you work with bacteria which usually have considerably smaller transcriptomes than the "common" model organisms that typical NGS guides are based on (fly, mouse, human etc...) you should rather take the lower end of these recommendations. Be sure that you rather sequence less deep and add as many biological replicates as possible to get robust statistics. Read length is typically of minor interest unless you aim to do some kind of transciptome assembly. For routine DE analysis a single-end 50 or 75bp run is often used. Paired-end has some advantages but if costs are limiting then as said rather use single-end and a bit lower depth and include a replicate more. I would see whether you can get quotes from facilities or companies. They often sequence on large machines such as the Novaseq where paired-end is standard and they typically guarantee results and minimum read depth since it is so standard these days. You simply send them RNA and the prep and sequence the samples. Might be more feasable than setting things up yourself if you have no experiece with NGS so far. Be sure that you talk to a bioinformatician before doing the experiments as proper experimental design depends on the analysis goal. For standard DE try to get at least three replicates per group.

[abhilashtripathi10]

Thank you for your suggestion. yes, because of budget constraints I am trying to go for 10-20million reads (50bp, SE)/sample. I have three biological replicates for each sample.