Hello,
I have ten samples of small-rna seq data to analyze but this is my first time working with small rna-seq data and I'm a bit lost with the fastQC quality control results after I remove the adapters. The small RNA is 50 bp single stranded and the quality report is good (mean QC30) before processing. However once I remove the adapters using fastx_toolkit Q30 and -l15 there is a dramatic drop in the quality after 30 nts and I lose roughly 60% of the data. I was wondering is this normal with short sequences? The adapters are of length 21 nucleotides so could the drop be because most of the sequences are shorter than 30 nucleotides? I know that I can trim these bases to 30 but I'm afraid that these processing steps have resulted in a loss of too much sequencing data.
Thanks in advance! Alex
I'm not entirely sure what fastx is doing internally, but I'd like to note that:
When you have insert size shorter than read length (as in small RNA), and if you have fairly low-quality data, then:
If you perform adapter-trimming only, you will enrich for low-quality tails because the high-quality tails will match adapter sequence and be removed, but the low-quality tails will have too many errors to match the adapter sequence and be retained. Thus it might appear that your overall quality was reduced as a result of the operation, but it wasn't. In that scenario, robust quality-trimming may be helpful.
Hi Brian, Thank you I think that makes sense! One query, which I apologize is very simple, when you say robust quality trimming do you mean trimming each read to 30 nt before the drop in quality begins or performing a different method altogether to remove the adaptors
By "robust quality trimming" I mean using a quality-trimming method that takes into account internal bases rather than the standard method of trimming bases until one is encountered that has a quality score above some threshold (which, I believe, is fastx's method). Note that quality-trimming and adapter-trimming are independent, and adapter-trimming should be done prior to quality-trimming. If you tell BBDuk to do adapter-trimming and quality-trimming it will do both at the same time, but for each individual read, quality-trimming will happen after adapter-trimming.
Thank you very much for your help, this is all a lot clearer now!
Did you use a specific small RNA kit to prep these libraries? If so, it must have come with instructions with the specific adapter sequence and if there is a need to selectively remove additional bases from from the front or end of the trimmed read (leaving the small RNA behind).
I didn't actually perform the library prep, that was already done . I am carrying out the analysis for my MSc project. I will check with the lab tomorrow though on their method!
Small RNA pipelines (e.g. miRquant ) take care of adapter removal etc. Were you planning to use one of these pipelines?
Something does not sound right though. After you trim data the quality should remain constant (since you removed low quality portions). Post images of the data before and after so we can see what is going on.
That's what I thought unless these sequences still have a mean score of Q30 which could be why there still present. Would that make sense? I'm hoping to first align to the HG19 and then also use mirdeep*
Thanks for your speedy reply!
Please use
ADD COMMENT/ADD REPLYwhen responding to existing posts to keep threads logically organized.If you feel that fastx_toolkit did not do a good job of trimming adapters then I recommend you to take a look at bbduk.sh from BBMap suite.