Weird FastQC reports on NextSeq 500 data

[nucleolus]

Greetings,

We recently obtained some NextSeq500 data and I've been tasked with processing it. We were provided with deconvoluted FastQ files. I ran those via FastQC (FastQC-0.11.3) and got some strange results on the 1) per base sequence quality and 2) per tile sequence quality (See attached). I've also attached the full FastQC report as a PDF.

Any thoughts - I don't know what to make out of these. Thanks,

J

[GenoMax]

There is nothing weird about these results. Looks like you have excellent sequence quality and there are no problems with any tiles. You appear to have a little adapter contamination but a pass through a trimming program (bbduk.sh, trimmomatic, cutadapt) can take care of that.

[Gouri Deshpande]

Base qualities look poor towards the 3' end

Hi,
I have much more weird looking plots for adapter content and for the base quality distribution


According to this fastqc result, looks like about 35% reads contain illumina universal adapters. Tried to run it through cutadapt. This flagged >45% adapters present.

Total read pairs processed: 4,605,709
Read 1 with adapter: 2,226,093 (48.3%)
Read 2 with adapter: 2,177,668 (47.3%)

[dpryan]

The quality, particularly for read #2 in a pair, typically decreases a bit with length, that's expected. Adapter contamination will happen whenever you read length is longer than the fragment you sequenced. When it's on the 3' end like that then that's what's happening. Just trim you data, as you should be doing anyway, and the results will be fine.

These plots aren't "weird", they're pretty much what I see all the time for many types of experiments. Don't worry too much when FastQC tells you a test fails, those aren't even applicable to many types of sequencing experiments.

[Gouri Deshpande]

Thanks dpryan