Problem with trimming ilumina adapter

Question

[silas008]

Hi. I'm trying trimming the adapters of my RNAseq and I found the adapter used, is the TruSeq Adapter, Index 1. But, when I use cutadapt or fastx_clipper the adapters remain in the sequences. I think that the size of the adapter make more complex the process of trimming.

I found too that the sequence of adapter is not the complete adapter but a part of that. That is:

TruSeq Adapter, Index 1

Complete = GATCGGAAGAGCACACGTCTGAACTCCAGTCACATCACGATCTCGTATGCCGTCTTCTGCTTG

Part = GAACTCCAGTCACATCACGATCTCGTATGCCGTCTTCTGCTTG

Maybe I should trimming the adapters using part of then or the correct would be trimming using the complete adapter sequence even if the trimming process is incomplete?

[dariober]

I usually pass to cutadapt only the first bases of the TruSeq adapter ('AGATCGGAAGAGC'), just like trim_galore. This way you capture several Illumina adapters in one go.

[silas008]

But, can I use one small part of adapter sequence?

[dariober]

Well, the straight answer is yes. cutadapt won't mind what sequence you use, of course. Question is, what do you expect to cut? cutadapt has several options about how to detect and trim. I'd suggest to do some experiments by picking or generating few reads which you expect to be trimmed and then play with cutadapt's parameters to see the results.

Answer 1

[GouthamAtla]

Why do you think the adapter remains after using cutadapt? There are very clear instructions for removing illumina TruSeq adapters in cutadapt documentation.

[silas008]

Ohh... thanks. I didn`t see this part of tutorial.

Answer 2

[Brian Bushnell]

If you are not sure what your adapter sequences are, and you have paired-end reads, you can determine them with BBMerge like this:

bbmerge.sh in=reads.fq outa=adapters.fa reads=100k

Then you can trim them with BBDuk, which is able to do a trim both by sequence-matching and overlap, resulting in greater sensitivity and specificity compared to algorithms that only use sequence matching.

(P.S. That command is for interleaved reads; for paired reads in 2 files, use the in1 and in2 flags).