Sequence biases toward read ends

[gavin.oliver]

I'm just having a look at the publicly available DH10B data with the FastQC program and notice some pretty huge spikes in sequence bias at the 3' end, with smaller rises at the 5` end. Being new to Ion Torrent data I'm wondering if anyone can explain what I am seeing?

Thanks in advance.

[SeqAA]

Do you mean slopes or spikes? Keep in mind the x axis is not linear with FastQC, so it tends to misrepresent 3' quality in PGM data. I think a few blogs have mentioned this.

I still use FastQC for a quick look at quality, but it's built around Illumina error models.