QC and biological repeats of small RNA sequencing data

Question

[kevolinchen]

When using fastqc to assess the quality of small RNA-seq data, what is a good results? (It's seems that the quality is not good for all the data)

How to deal with biological repeats (e.g. 3 fastq files)?

Is it ok to merge them first? If so, how to merge them, any tools? Can I just use

cat A_1_R1.fastq A_2_R1.fastq A_1_3_R1.fastq > A_merge.fastq?

What is the difference between sRNA-seq and RNA-seq when filtering?

For now I just try to trim the adapter and remove reads form repeat region of genome.

I think really need some tutorial for sRNA-seq anaylsis....

Do any biostarers have some tools or pipelines to share for sRNA-seq analysis?

Thank you~

Answer 1

[Devon Ryan]

Take all of the warnings/errors that FastQC produces with a very large grain of salt (they're most appropriate for DNA sequencing).

Regarding the biological replicates, keep them separate.

sRNA-seq has a different library prep. than regular RNAseq, where there's a stringent size-selection step in the former (resulting in mostly small (<35bases or so) transcripts being sequenced). Regular RNAseq doesn't normally have a size selection step (excluding the fact that some kits won't efficiently produce transcripts less than ~200 bases), though either poly-A selection or ribo-depletion are common.

For tutorials, search pubmed.

Regarding pipelines, it depends on whether you're mostly interested in miRNAs or another single species or if you want everything in a single go. Most of the published pipelines (e.g., miRdeep2) are targeted toward a single RNA species...though since most people are interested in miRNAs this makes sense.

[kevolinchen]

Thank you. But how to integrate biological replicates? Average after counting or ...?

I am now trying miRdeep2.:)

[Devon Ryan]

You would use them in the downstream statistics, which would rely on replicates due to using a negative-binomial model.

[matt.arno]

Is there any way to view samples together on the same chart/graph in FastQC, or any other fastq QC analysis package?

Cheers,
Matt

Answer 2

[Lars]

You could give DARIO a try.

[kevolinchen]

Thank you, that seems to be a user friendly web tool~ But I am afraid that my data are too big to analysis online (44 samples). So, I'd better find a pipeline or build a pipeline can be run on our server.