Weird overrepresented K-mers in FastQC

[EcoRInya]

Hi! We have recently done the first run on a new NextSeq 500 machine which produced reads with weird FastQC results (attaching below).

This is a paired end sequencing of MNase-seq sample. Libraries were prepared using NEBNext Ultra kit using TruseqLT Illumina adapters. Sequencing was done using NextSeq 500 sequencing kit v2 high output.

I don't have a lot of experience with sequencing so far but FastQC results like these look worrying for me.

Could please anyone explain to me what are these numerous k-mers overrepresented along the reads, including the center of the read (which means that is it is unlikely to be untrimmed adapters)? Is this a problem with library prep or with the actual sequencing on the machine?

[GenoMax]

I am not familiar with MNase-seq but perhaps that enriches some sequence that may be leading to that particular enrichment of k-mers.

Different criteria (pass/fail) used by FastQC are settings that can be changed and are configured (default) for normal genomic sequencing. Any experiment that deviates from the norm is likely to generate some fail flags. That does not automatically make your data bad. You should keep those observations in mind as you proceed with your analysis. If you are unable to get reasonable alignments then you can circle back and check if these k-mers are responsible for it.

The string of poly G's is a characteristic of NextSeq data. NextSeq uses a 2-color chemistry so absence of any signal is considered a G. See this link. You can trim those reads out using GGGGG as seed for the trimming program.

[EcoRInya]

Thanks, GenoMax! I guess we can test it the next time when we sequence a different type of libraries (like from cross-linked ChIP) and look if we still see these weird k-mers. The prediction is that we shouldn't. I didn't know about GGGGG in NextSeq data. Thank you for pointing this out!