Duplicate Sequence/High Overall Kmer Content

[lala2013]

I am learning to interpret fastQC reports. I combined all the fastq files for R1 and R2 and ran fastQC on them separately. The rest of the summary look good except the duplication levels and Kmer content. For both R1 and R2, there is a slight increase of duplication levels after 10 (I know this is kind of to be expected). What bothers me is the overall high kmer content (above 70%) across all positions in read. I do not have any explanation for it, except that it looks like a systematic error. Is there anything I could do (trimming, filtering, etc.) about it or should I just leave it like that?

Thank you very much!

[kmcarr]

I am learning to interpret fastQC reports. I combined all the fastq files for R1 and R2 and ran fastQC on them separately. The rest of the summary look good except the duplication levels and Kmer content. For both R1 and R2, there is a slight increase of duplication levels after 10 (I know this is kind of to be expected). What bothers me is the overall high kmer content (above 70%) across all positions in read. I do not have any explanation for it, except that it looks like a systematic error. Is there anything I could do (trimming, filtering, etc.) about it or should I just leave it like that?

Thank you very much!

Those over represented Kmer plots are difficult to interpret and often leave an impression that things are far worse than they really are. First of all, the numbers on the Y-axis are NOT %, they are relative amounts. In your graph the AAAAA kmer is most over represented at position 1 of the read so that level of over representation is defined as '100'. The level of AAAAA over representation at other positions are then plotted relative to the over representation at position 1. In other words the high point for ANY kmer plotted in this graph will always equal 100.

The more important information is in the table below the graph. What is the actual level of over representation of the AAAAA kmer, the Observed/Expected ratio reported in the table?

[lala2013]

It actually makes more sense after reading your reply! Below are the two tables after Kmer content graphs.

For R1:
Sequence Count Obs/Exp Overall Obs/Exp Max Max Obs/ExpPosition
AAAAA 19020565 3.0625381 4.010783 1
TTTTT 19299775 3.0321934 3.3969162 10-14

For R2:
Sequence Count Obs/Exp Overall Obs/Exp Max Max Obs/ExpPosition
AAAAA 18148705 3.1093986 4.3404694 1

For both R1 and R2, the most observed AAAAA is at position 1. Does this say something about the data? But the most observed TTTTT is at 10-14 position. Also, assuming the count means the number of time the kmer is found in my data, is ~20million a big number? This is exome seq data

[kmcarr]

Personally I don't see anything in your QC that would worry me. The AAAAA & TTTTT over represented kmers are probably just some sequencing artifacts and really are not very abundant. Don't waste time stressing about them.

[lala2013]

Thank you!