miRNA seq suggested read count and applicability on Illumina iSeq 100

Question

[ivanj]

Dear all, I am thinking about to perform an experiment on miRNA expression. However, the data from the literature about the optimal number of reads for expression experiment is quite variable. In some sources it was stated that 100.000 - 1M reads is enough. Others state 2M is suggested. I would like to know, based on your experience, what would be the best tradeoff by taking into account the instruments max number of reads which is 4M for single end sequencing? Also, what is the expected percent of mapped reads in miRNA expression experiment (I guess this percent should be counted for those 100.000 - 1M, not the raw reads)? The standard Illumina library prep protocol should be used on human samples. Thank you in advance, Ivan

[Buffo]

It depends of your specie of interest and expected miRNAs to calculate the coverage needed. And also of what you want to analyze, , for example, to analyze expression (in mouse) 20M would be enough, However, to distinguish between isomirs and/or look for new probable miRNAs you will need more coverage about 80-100M.

[ivanj]

I am planning to analyze miRNA expression in human samples (and not the discovery, just the expression). I think that 20M reads per human sample would be way to high.

[h.mon]

If you want to analyse just differential expression of miRNA ignoring isomiRs, then 1-2 million reads should be enough.

What Buffo is pointing at, though, is you will be missing a lot of information if you follow this route, because there is extensive variation in the isomiRs produced, and they may have different roles / targets on the cell. I am not up-to-date with the literature on isomiRs, but these two papers will help you get started:

Evaluation of high-throughput isomiR identification tools: illuminating the early isomiRome of Tribolium castaneum

A Comprehensive Approach to Sequence-oriented IsomiR annotation (CASMIR): demonstration with IsomiR profiling in colorectal neoplasia

edit: the following paper has a discussion on sequencing depth and isomiR detection power:

CAP-miRSeq: a comprehensive analysis pipeline for microRNA sequencing data

And this one does the same, but it seems it groups isomiRs under one miRNA:

Assessment of microRNA differential expression and detection in multiplexed small RNA sequencing data

[Buffo]

Keeping in mind that there is no method to sequence just miRNAs and exclude other kind of short RNAs, if you consider 20M too high for human... go ahead, personally I do not think so.