Adapters trimming about afterqc HOT 10 OPEN

How many adapters got removed, and how many not?

AfterQC trims adapters for pair-end sequencing data based on searching for best overlap. But if there are too many sequencing errors, it will make AfterQC failed to find the overlap and consequently failed to trim adapters.

Recording the adapter information in the report is a good idea, I will implement it in future release.

from afterqc.

multiqc report before any processing

multiqc report after AfterQC

AfterQC report of one sample

Also quick question, what would happen if I run AfterQC again on good reads?

from afterqc.

If you run AfterQC with good reads, you should get every good read being kept.

But you can have a try.

from afterqc.

From the figures you post, seems that the adapters those cannot be trimmed are at positions < 40, which means these DNA templates are shorter than 40?

from afterqc.

Yes, which is weird. The original paper says:

KneadData. KneadData incorporates Trimmomatic and bowtie2 for filtering and human sequence removal, respectively. Reads were scanned with a four-base wide sliding window and trimmed when the average base Phred score drops below 20. Trimmed reads shorter than 70 nt were discarded.

So, I'm not sure if they trimmed adapters.

For me it is important to understand how AfterQC works, e.g. if what I got is expected, because of many sequencing errors, which makes AfterQC fail to find the overlap, and if it is advised to apply additional trimming after AfterQC in such case.

from afterqc.

It's not the major reason that sequencing error results in overlap detection failure, I think less than 1%.

From the figure you post, the major reason is that the DNA templates are too short. AfterQC requires at least 30bp overlap to detect adapters, that's why the adapters under or near position 30 are not trimmed but the adapters beyond position are all trimmed.

I am very curious why your sequencing library contains so many short fragments, I suspect that most of them may be self-ligased adapters. Did you perform fragment length selection after adapter ligation?

from afterqc.

I'm curious myself. It's not my article, I'm analyzing public dataset.

Here is the article. Please, read the section "Shotgun library sequencing and quality control". But regarding sequencing protocol they reference one more article, so I'm unable to answer your question currently, need to read more.

Additionally, here is sequence quality after AfterQC (raw is similar, but worse):

from afterqc.

Ok, seems the quality drop cycle by cycle is a bit serious, and there are two gaps...

from afterqc.

Yes, I thought that the gaps are bubble related. That's how I found AfterQC, because I googled the tool, which can correct that. But AfterQC didn't find anything. So I have no idea how those drops are generated.

from afterqc.

UPDATE. For NexteraXT transposon protocol when the insert is shorter than the length of a single read it results in the presence of adapter at the end of the read. Article, blog post, biostars post.

But for metagenomics datasets I don't think we can identify read pairs with small insert sizes prior. I will try to play with fastqc report by increasing k-mer size and see if I can identify the issue by looking on k-mers (i.e. without knowing nextera adapters sequence).

I think AfterQC can deal with this particular problem, but for that you either need to check against particular sequence or change the trimming algorithm somehow. But anyway just having adapters-related qc report would be great. I don't think that the tool should cover all the possible situations, but it should provide the user with information to see potential problems.

from afterqc.