Is It Necessary To Remove Adapters In All Orientations When Preprocessing Ngs Data?

Question: Is It Necessary To Remove Adapters In All Orientations When Preprocessing Ngs Data?

7.3 years ago by

JacobS • 940

Cleveland, Ohio

JacobS • 940 wrote:

I'm prepping some NGS Illumina data for downstream analysis. To begin, I want to remove any sequencing/ligating adapters and multiplexing (barcoding) tags. To do this, I am using fastx_clipper, which is part of the FASTX-Toolkit. I've also using Trimmomatic for this in the past.

Example command usage: fastx_clipper -Q33 -a TGGAATTCTCGGGTGCCAAGGAACTCCA-mid_tag_insert-AATCTCGTATGCCGTCTTCTGCTTG -l 14 -M 7 -i Input.fastq -o Output.fastq -v -c

Here is my question... Both of these software packages only scan for a single orientation of the adapter you provide within the Illumina reads. However, I find many sequences in all orientations of the adapter, namely: forward, reverse, forward complement, reverse complement. In the forward orientation, the software detects and trims the adapter in >90% of the reads, but in the other 3 orientations the software only detects are trims adapters in ~5% of the reads.

So, is it possible for the adapters to be found in different orientations than the forward sense, or am I seeing artifacts of non-strict adapter matching? Do people usually trim adapters in every possible orientation? Any other suggestions for successfully handling adapters?

Thanks!

trimming ngs filtering • 6.6k views

ADD COMMENT • link •

modified 7.3 years ago by Manu Prestat ♦ 4.0k • written 7.3 years ago by JacobS • 940

7.3 years ago by

Istvan Albert ♦♦ 86k

University Park, USA

Istvan Albert ♦♦ 86k wrote:

Seeing an adapter in the forward orientation is the result of a DNA fragment being shorter than the read length, it is a "normal" occurrence in these cases. The Illumina TrueSeq indexed sequencing adapters were designed in such a way that the same adapter sequence will be found on reads coming from both strands.

In that case adapters present in any other orientation most likely indicate a protocol failure, in which case probably the entire read should be removed.

ADD COMMENT • link modified 7.3 years ago • written 7.3 years ago by Istvan Albert ♦♦ 86k

@Istvan Thanks for the answer. This is what I expected, which makes me unsure of the FASTX results. Perhaps my stringency is simply too loose, only requiring 7 sequential adapter bases to be matched. I'll try a few variations and report back later.

ADD REPLY • link modified 7.3 years ago • written 7.3 years ago by JacobS • 940

also to correct myself some (not all) Illumina adapters are designed to produce the same sequence on both strands

ADD REPLY • link written 7.3 years ago by Istvan Albert ♦♦ 86k

Can you please specify which ones do and which ones don't? Thanks!

ADD REPLY • link written 6.0 years ago by enricoferrero • 810

Similar posts • Search »