High T and C at end of reads

[illinu]

Hi,

I recently ordered sequencing of two libraries (mate pair and paired end) with an Illumina MiSeq (reads of 301 bp). Both libraries contain high amounts of T and C after base 250 in forward and reverse reads (picture attached).

To my understanding this is due to sequencing errors and I should trim of all bases after position 250?


Thank you for your advise

[dpryan]

It actually sort of depends on what you're trying to do and which aligner you intend to use. It's likely that the quality has decreased in the biased region and many aligners may simply soft-clip those regions.

[illinu]

I am going to do de novo assembly. I will firts try with AllpathsLG maybe Soapdenovo2 as well.

[kmcarr]

Hi,

I recently ordered sequencing of two libraries (mate pair and paired end) with an Illumina MiSeq (reads of 301 bp). Both libraries contain high amounts of T and C after base 250 in forward and reverse reads (picture attached).

To my understanding this is due to sequencing errors and I should trim of all bases after position 250?


Thank you for your advise

Illinu,

Don't base decisions about trimming your reads off of these plots from FastQC. They are good for getting a global impression of your data quality but don't give you information about individual reads. Another thing to keep in mind is that those diverging lines at the very end of the plot are simply an artifact of the way FastQC groups data for plots of long reads; the x-axis is not uniformly scaled. The farther you get from the 5' end of the read the larger the window FastQC uses to bin data. At the 3' end the plot points are actually averages over 50bp windows. Your reads look fine up until, in your case, it gets to the very last window where it only has bases 300-301 left. Those last, diverging points are data for just the last two bases of your reads.

Don't blindly trim off huge chunks of every read. Use dedicated NGS read trimming tools (e.g. trimmomatic or trim_galore) which examine reads individually and trim each appropriately. These tools can detect and trim adapter sequences as well as poor quality bases.

[SNPsaurus]

Are the reads running into the adapter at the end? That could cause a shift in GC content.

[illinu]

Thank you all for your replies. This helps. I looked for adapter contamination and there is none. I was using trimmomatic to trim the last 50 bases but it is true that it's too much. I ran allpathslg and the first result is pretty bad probably due to the trimming since allpaths needs reads to overlap. I will try a second run with the entire reads.