Hello,
We recently changed our data provider, and I’m noticing some unusually high-quality scores in certain blocks. Could this indicate a problem with my data? I’m currently running the following commands:
Should I modify anything in these commands? Alternatively, should I use FastQC to assess the read quality first, and then running the previous command in order to generate the demux.qza used in denoise step after.
It's most likely not an issue with your data — are you using NovaSeq or iSeq data by any chance? See here for some related discussion:
Some newer Illumina sequencers bin quality scores into 4 bins to reduce the data footprint, which will produce an output like you're seeing above. See the topic above and here for more details.
Thank you very much for your answer.
I had already seen them before publishing my issue, but I didn't really understand what I should do in this case. (Sorry, I'm still a beginner in this field.). could you guide me more please !!
"Should I modify anything in these commands? Alternatively, should I use FastQC to assess the read quality first, and then running the previous command in order to generate the demux.qza used in denoise step after."
Nothing is wrong with your commands - the interactive quality plot looks as expected for sequencing data that's been quality score binned (NovaSeq and iSeq both do this).
You're welcome to cross check the read quality in FastQC, but one thing I'll point out about using FastQC is that the x axis on the plots it generates can be a little confusing if you're directly comparing to q2-demux's summarize visualization, because FastQC actually performs a separate binning of the nt positions:
So in particular, you're probably looking at a psuedo-log type of plot, where only ~50 or so box plots are displayed, rather than the ~290 box plots that you shared in the screenshot above. Just something worth keeping in mind!
In either case though, you don't need to modify anything you've done up to this point - you're now just using this quality plot to assess the filtering params you'll set when denoising.
Thank you very much for all these clarifications. As far as I understand, QIIME2 only supports sequencing data from Illumina MiSeq when plotting graphs for the first step of quality control. It does not support NovaSeq or iSeq. For this reason, we must use other tools such as FastQC and MultiQC to visualize forward and reverse reads and to properly set the parameters for the denoising step.
Do you have any other suggestions for tools that can generate boxplots similar to those produced by QIIME2?
When running MultiQC, I see the graphs for forward and reverse reads combined in the same plot. Is there another tool that separates them, or do I need to run MultiQC separately on forward and reverse reads to view them individually?
QIIME 2 does support visualization of different types of sequencing data such as NovaSeq or iSeq - the box plots just look different than with MiSeq data since the quality scores have been binned.
If you haven't already, I'd recommend checking out the link I shared in my previous response (FastQC compared to QIIME 2 quality plot) - you can see an example of the visual difference between the two plots (even though this example is from a different type of sequencing data).
I can't speak to the usage of MultiQC, as this isn't a tool I've used before - but you're welcome to cross reference the outputs from QIIME 2 with other tools to compare results.
I use MultiQC before importing my reads into QIIME 2. You can use the toolbox (right side of the window) to hide or show forward and reverse reads using the distinctive part of the filename:
