Fastqc Per sequence quality scores - Two peaks?

Wilro · 05-20-2017, 04:08 AM

Hi,

I am struggling to find possible explanations for seeing two distinct peaks in my "per sequence quality scores". I am very new to this, so hope it's not a very silly question. As far as I understand it should not be a major concern, however I would like to understand why this is.

There is also two peaks for GC content in sequences, with very low adapter content at the end of reads.

Sanger/Illumina 1.9 encoding
The sequence data is from an unknown clinical sample from hospital, for the purpose of identifying possible pathogens.

Kind regards,
Wilro

mastal · 05-20-2017, 04:58 AM

I wouldn't worry about the two peaks in the per sequence quality scores although it looks a bit unusual.

Are the reads all the same length if it is Illumina data, or have they already been trimmed by the sequencing center?

The per sequence GC content is more important, and could indicate contamination, but if you are expecting pathogens in the sample, then it could indicate a mixture of human genome and pathogen genomes perhaps.

Wilro · 05-20-2017, 06:26 AM

Reads are all the same length and have not yet been trimmed. Downstream analysis of the data revealed multiple pathogen blast matches, as well as contigs matching to human reference genomes. So your explanation would make perfect sense.

Thanks for your quick reply

tirohia · 05-20-2017, 05:18 PM

I was about to post on exactly the same topic - I've got two peaks, one quite sharp in the GC content across all 12 samples in an experiment. As best I can tell from googling is that contamination is a likely cause.

The group that sequenced the samples for us though suggest that it may have been caused by the rRNA depletion step on initially poor quality RNA (our samples were borderline, quality -wise going into this). I'm wondering how likely others might think this to be? Or how that would work?

It's also been suggested that I map the RNA back to the Grch38 and filter, retaining only reads that map back to known genes. Which strikes me as ... dubious, given that if it is contamination, I don't know what the contaminant is.

Also, if it is contamination, how does one deal with that bioinformatically? Obviously it's going to reduce the validity of any results coming out the other end, but it'd be nice to not have to right off the whole thing.

nucacidhunter · 05-21-2017, 03:44 AM

GC content double peak in RNA-seq does not automatically imply contamination and it is common. Contamination or residual rRNA presence can be checked by BLAST or mapping. Posting the whole FastQC plots will be useful to diagnose any potential issues.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Deleting sequence lacking quality scores from FASTQ?	bwc22	Bioinformatics	2	02-23-2016 09:59 PM
a question about FastQC's quality scores	kshen	Bioinformatics	3	03-27-2013 02:56 AM
FASTQC: Quality Scores	rlowe	Bioinformatics	2	05-02-2012 01:37 AM
Per Base Quality scores in FastQC	mittymat	Illumina/Solexa	3	03-30-2012 06:34 AM
Quality scores and sequence alignment	warrenemmett	General	0	09-01-2010 06:35 AM

05-20-2017, 04:58 AM	#2
mastal Senior Member Location: uk Join Date: Mar 2009 Posts: 667	I wouldn't worry about the two peaks in the per sequence quality scores although it looks a bit unusual. Are the reads all the same length if it is Illumina data, or have they already been trimmed by the sequencing center? The per sequence GC content is more important, and could indicate contamination, but if you are expecting pathogens in the sample, then it could indicate a mixture of human genome and pathogen genomes perhaps.

05-20-2017, 06:26 AM	#3
Wilro Junior Member Location: Stellenbosch, South-Africa Join Date: Mar 2015 Posts: 2	Reads are all the same length and have not yet been trimmed. Downstream analysis of the data revealed multiple pathogen blast matches, as well as contigs matching to human reference genomes. So your explanation would make perfect sense. Thanks for your quick reply

05-20-2017, 05:18 PM	#4
tirohia Member Location: Auckland, NZ Join Date: Nov 2011 Posts: 46	I was about to post on exactly the same topic - I've got two peaks, one quite sharp in the GC content across all 12 samples in an experiment. As best I can tell from googling is that contamination is a likely cause. The group that sequenced the samples for us though suggest that it may have been caused by the rRNA depletion step on initially poor quality RNA (our samples were borderline, quality -wise going into this). I'm wondering how likely others might think this to be? Or how that would work? It's also been suggested that I map the RNA back to the Grch38 and filter, retaining only reads that map back to known genes. Which strikes me as ... dubious, given that if it is contamination, I don't know what the contaminant is. Also, if it is contamination, how does one deal with that bioinformatically? Obviously it's going to reduce the validity of any results coming out the other end, but it'd be nice to not have to right off the whole thing.

05-21-2017, 03:44 AM	#5
nucacidhunter Jafar Jabbari Location: Melbourne Join Date: Jan 2013 Posts: 1,238	GC content double peak in RNA-seq does not automatically imply contamination and it is common. Contamination or residual rRNA presence can be checked by BLAST or mapping. Posting the whole FastQC plots will be useful to diagnose any potential issues.