[Bug]: vsearch does not fit for adapter search

[defendant602]

What happened?

Vsearch tool seems not to fit for adapter search in a read, as it uses a global alignment method. If there are two or more adapter1 sequences in the read(concatemer), vsearch will only report one hit. In contrast, local alignment tools will report all the hits.

Operating System

ubuntu 18.04

Workflow Execution

EPI2ME Labs desktop application

Workflow Execution - EPI2ME Labs Versions

all

Workflow Execution - CLI Execution Profile

Docker

Workflow Version

all

Relevant log output

None.

[nrhorner]

Hi @defendant602

Thanks for your question, and apologies for the very late response. We will respond to your question shortly.

[syyyChen]

Hello,

I've also encountered this issue where approximately 5% to 10% of valuable molecules seem to be mistakenly discarded in my sequencing data. I noticed this problem was mentioned in issue#530 of vsearch. However, it appears the issue hasn't been addressed yet.
Do you have recommendations for alternative tools that can be used for adapter searching without this problem?

[cjw85]

@syyyChen If you are observing the effect described in torognes/vsearch#530, then the workflow discarding such molecules would be the correct behaviour. Such molecules could be split and useable data rescued if as @defendant602 correctly pointed out we were using a tools that reporting all hits.

As it stands the use of vsearch limits the code to identifying and correcting only certain classes of chimeric molecules. We are reviewing the methods used in this step.