How many reads is sufficient for differential expression analysis in Small RNA Seq? and why?

Question

[Aan]

Hello everyone,

Nice to meet you all.

I am a newbie in sequencing technology. I have done sequencing samples with small RNA seq using MiSeq. and several issues appeared. A lot of samples produce very low reads (ranging from <100 reads up to 1.2 million reads, mostly less than 1 million reads). Illumina staff said that for differential expression analysis using small RNA Seq, recommended number of mapped reads should be between 1 M - 5 M (based from other studies). I also searched for other studies of small RNA seq, and they used HiSeq with mapped reads more than 1 million. I plan to resequencing my samples using HiSeq, but there are few questions.

1. Perhaps someone know why the recommended number of reads for differential expression analysis with small RNA seq is between 1 M-5 M?? my supervisor wanted to know the answer to this question. I am not able to find answer on this question.
2. i analysed my fastqfile data with third party program called strandNGS. Supposed i resequencing several of my samples with HiSeq (especially those reads less than 1 M), is it okay/advisable to combine analysis of HiSeq fastqfiles and MiSeq fastqfiles with this third party program (or any other program)?

Thank you very much for your attention.

Regards.

[GenoMax]

1. Have you looked at other published studies to see what people have done?
2. If you are going to re-sequence on HiSeq then that alone should generate enough data for you to not to have to worry about using MiSeq data. If you are planning to re-sequence the same libraries on HiSeq then re-pool them (amounts) based on MiSeq results to ensure more or less even numbers from HiSeq run.

[Aan]

Dear genomax, thank you for the response.

1. I read few studies, especially one study that reported the correlation of number of reads with how well that number represents distribution of miRNA in sample was by Metpally et al in their 2013 study (comparison of analysis tools for miRNA high throughput sequencing using nerve crush as a model). in the study, it was suggested that the correlation of number of reads and how well it represented the sample was fairly stable if the mapped reads were from 1 million or 1.5 million reads. But thank you very much for the suggestion. I will present other studies regarding this matter to my supervisor.
2. Thank you very much for the advice on HiSeq samples.

Regards.