FastQC analyses of trimmed MiSeq reads kmer content

[akjones]

Hello all,

After doing quite a bit of "lurking" with regards to other posts on this topic I have found that although my problem/question is similar to others out there, it is not the same so I have decided to post my own thread.

I have Illumina MiSeq data for a bacterial isolate we know to have high GC content. I analyzed raw data, and trimmed datasets with fastQC. Most areas of the analysis look good with the exception of per base sequence content, sequence duplication levels, and kmer content. Although these areas improved with aggressive trimming, they still do not receive a "pass". I have been doing all of my trimming with Trimmomatic. Aggressive adapter/quality trimming greatly decreases the over represented k-mers, but does not do much to help the per base sequence content. A hard trim of 10bp from the 5' end of the read greatly improves the per base sequence content, but does not do much to improve the overrepresented k-mers.

As I am using this data for a de novo assembly I am really loathe to do a flat 10bp crop because I would like to keep as much data as possible and as many long reads as possible.

I am attaching the reports for the stringent adapter quality trim, and 10bp crop data sets. Any and all input would be appreciated. If you would like to see the raw read assessments as well, let me know and I will provide those.

Thanks,

~Ana

[akjones]

I couldn't fit all attachments, here are the 4 for the stringent trim. R1 and R2 are from the left and right, respectively.

~Ana

[mastal]

Hi,

I had a look at the Per base sequence content plots and they don't look too bad.

You could try doing de novo assembly with and without the 10 base pair crop at the 5' end, and see which gives better results.


As for the 3' end, it's usually just the last base that has a much lower quality than the rest of the sequence, so just trimming that should remove the effect you see at the very end.

[akjones]

Thanks mastal, it's reassuring to hear they don't look too bad. My thoughts were similar with regards to the assembly. Do you know if there is a way to trim a specified number of bases from the 3' end using Trimmomatic? As far as I know the "CROP" function only lets you set how many bases you want to keep, and since reads are of different lengths my understanding is that this can result in an over trim of high quality, longer reads and no trim to shorter reads.

Thanks again,

~Ana

[mastal]

As an example of how CROP works in Trimmomatic, if you have reads of 151 bp long, and you want to remove the 151st base, you do CROP:150,
and as you said, if some of your reads are shorter, they won't be affected.