Kmer overrepressented in an exome analysis

[jpofmars]

Hi,

I have a problem with an exome analysis. The sample was sequenced with illumina HiSeq 2000 in paired-end with 100 cycles.
When I run FASTQC to see the quality of my samples, I have kmers which are overrepresented. (TTTTT, AAAAA, GGCGG). I join the picture of kmer content.

I have poor feedback on this kind of analysis, whereas it is the first sample (for exome) where I see that.

Do you think there was a problem during the capture of exome ? Or is it a specification of my sample ?

Thanks.

[jpofmars]

For information, I perform the alignment and I see if all the exoms were covered with bedtools.
The bedfile of the exom was genereted with ucsc.

On 389394 exoms 244372 have a coverage of 0.

[amitm]

Trim bases from 5'-end

hi jpofmars,
Though I haven't done an Exome analysis, but I have seen the same issue with RNA-Seq data. As per the comment posted by BabrahamBioinf, creator of FastQC (@ YouTube page of FastQC tutorial vid -> link)

The random hexamers used during priming the library are not "that random" and hence u see the high peaks of k-mer at the 5' end for initial 10 bases or so. At the 3' end we use oligo-dT.

In my experience (RNA-Seq), trimming the first 3 bases from 5' end and trimming a couple of bad quality bases from the 3' end, majorly reduces the k-mer occurrence.

Attached are k-mer profile plots before and after trimming.