Heys,
I'm working with WGS data and after running the fastqc I detected that my samples still have adapters. I succesfully removed them with AdapterRemoval and Cutadapt, but I was wondering how I could compare the different softwares. I know that adapter removal's software is a sensitive topic but as I just have two samples, I would like to compare the results I obtain with the three softwares (including Trimmomatic). How could I do that? I'm just thinking in comparing total reads, but a part of that what else could I do?
Thanks in advance, Gabriel
1 answer
Don't overthink the adaptor removal.
The reason people prefer certain software often has to do with additional features that the software has beyond just the adapter removal.
For example if you wanted to remove all reads that have the adaptor on them and keep reads with no adaptor, then only cutadapt can do it, trimmomatic cannot.
If you wanted to process interleaved FASTQ file only AdapterRemoval can do it, and so on.
Each software offers features beyond the original use case. If you "only" need to remove adapters each of these should work similarly well.
Log in to answer this question.
-
merge paired-end sequencing
written by gubrins
-
Which adapters to use for trimming?
written by whb •
-
Filter per tile quality
written by gubrins
-
Trimming P5 and P7 dual index adapters (and other QC)
written by willnotburn •
-
Clipping of KAPA adapters
written by mewgia •
-
Short Insert size and aggressive trimming
written by Sharon
-
How to remove adaptors if you don't know their names/sequences? (eg from public data)
written by rioualen
-
Adapter Removal regardless posistion
written by biofalconch
-
How to remove adapters from DNA-seq illumina reads?
written by hed.robin •
-
How to deal with adapter dimers?
written by BioinfoNovice
Is that so? I would (in fact I do) simply use cutadapt, well-tested and used by hundreds of labs for almost a decade now. Don't overthink simple things like adapter trimming. Check with fastqc before and after trimming and if the tool managed to remove adapters then go on with analysis. For comparison better see whether there are papers doing that since proper comparison can be tedious. Both trimmo and cutadapt are fine.
Thanks for the quick reply. I said that sentence because when checking in the internet I found several users recommending several softwares (cutadapt, adapterremoval, trimmomatic, bbduk, ...). ATpoint, do you also do quality trimming? Thanks!
No, I do not explicitely trim qualities but I also am not into assembly, so if you vall variants or assemble genome it might be necessary even though most tools should probably be able to simply ignore bases below a certain quality. Most variant callers do, no clue about assemblers.
I recommend
bbduk.sh. It is one of the few packages that eliminates residual contaminants by overlapping R1/R2 read ends so it can remove that last base or two that may remain.That said, most aligners should take care of adapter contamination during alignment by soft-clipping the part of read that does not align so strictly speaking you can get away with not doing trimming. Adapter stretches generally get poor Q-scores and these should get eliminated as a part of this.
If you are planning to do any
de novowork the trimming should be the norm to remove all extraneous sequences. Quality based trimming should only be needed for this analysis.Heys again,
last question, what's the difference between trimming and marking the adapters? I trimmed them with cutadapt but I wanted to implement GATK's pipeline and I've seen they have a function to mark the Illumina adapters and assign them a low quality score, removing their contribution to read alignment and alignment scoring metrics. Both are ways to deal with adapters I imagine, which one is the best or it depends on you?
Thanks for your help!
Do not add new questions as answers to existing questions. Ask in a new post instead.
I'm moving this post from an answer to a comment.
Trimming is removing the affected bases from your original data before any alignments are done. Marking adapters is done with pre-aligned (one assumes non-trimmed) data files. I like to trim my data so I know all extraneous sequences are gone (or not present). Only adds a small amount of time (
bbduk.shis very fast) and I don't need to think about it again.Thank you very much for your comment, really nice to have inputs from other people :)