interpretation of FASTQC Overrepresented Kmers

[mattanswers]

I am analyzing our results of 91 base single end Illumina sequencing with FASTQC and have attached two .png images, one of the graph and the other of the graph and list of kmers. The sample, part of a ChIP-Seq experiment, is 'input' DNA.

I have a couple of questions with regard to interpretation.

First, a large portion of the listed kmers are part of the adaptor:
GATCGGAAGAGCTCGTATG. All kmers in postions 1-9 are part of the adaptor, but only the first 13 bases of the adaptor. Kmers covering bases 14-19 of the adaptor are listed, but their positions are listed as 85-86 ! It seems like sequences were inserted inside the adaptor ? Does anyone have an explanation for this ?

Second, in the graph of relative enrichment vs position in read, there is gradual rise in the 6 listed kmers from positions 30-34 to about position 70 or so. Because 5 of these 6 kmers included in the graph are part of the adaptor, as can discerned from the leftmost part of the graph in postions 1-5, what does it mean that the relative enrichment of these kmers rises from position 30 to position 70 ?

[Yilong Li]

Hey,

we've seen a similar rise of adaptor kmers towards the ends of the sequences. We haven't done anything formal analysis, but since we got paired-end sequences, we've been able to align the paired reads together, and it seems that many reads having adapter kmers originate from DNA fragments that are shorter than the read length. When this happens, sequencing will first proceed through your original DNA fragment and then continue to sequence the adapter sequence located immediately after it.

We have also seen weird patterns in nucleotide distributions in <8 bps of the 5'-end of the reads, but have no idea where it comes from. If you find out let me know.