Hi all,
Here are my results for quality plots when QIIME2/2018.11 and QIIME2/2019.01 are used. Is it because of random sampling?
2018
2019
Yes!
To clarify, the results aren't different because you used two different releases of QIIME 2 - you would see the same kind of differences by running the command multiple times within the same release of QIIME 2.
Hope that helps!
1 Like
Yes, it makes sense. Then using the plots for making decisions on truncation will be a little deceiving, right? Because my previous plot showed that everything was above 20 but now my reverse read sequences after position 245 are below 20, so I would remove them now but had not remove anything before.
1 Like
Hi @Negin!
I missed something earlier: your read lengths are significantly different between your first set of images and second set of images. This represents some change in upstream processing, or perhaps you swapped out data files on accident? The easier way to confirm is to load these visualizations, and navigate to the import nodes on the provenance tab:
Your provenance tab might look different, it really just depends on what steps you performed before running demux summarize. Specifically you will want to look at the node with the "import" action. In that, you will see one or more files listed, with md5sums (this will also vary depending on how the data was imported). If you compare these numbers between your two visualization, they should be the same (but I think that you will find that they aren't, which means they are different data).
Keep us posted!
1 Like
Actually, what I did was to re-visualize the qza file that was imported using an older version of QIIME2, so the second file does not show me the processes and steps taken. Could that be the problem? I haven’t imported using different versions but visualized using different versions so should have looked the same.
They don’t have different length though. The image is cut. Both are 250 nts. The problem is that I just visualized them using different QIIME versions but not imported twice so why would it look different?
Have you actually looked, though? Provenance will always be there... Feel free to share the files here (or send me a link to download them at) if you want a second set of eyes.
Hmm, it really looks like its not just a truncated image... Please share the QZV files, besides the quality summary plot there are several other useful plots (and provenance) that are very helpful, and will hopefully let us sort out how you have gotten to this point.
Thanks!
Sure. Thanks. Here are the files. Yes, I looked, the 2019 file does not show details of the provenance as the original does.
2018
paired-end-demux.qzv (289.9 KB)
2019
paired-end-demux.qzv (290.3 KB)
Great, thanks!
It's right here:
And here is the 2018:
The execution uuid is the same in each QZV's provenance tree, which is good confirmation that these are the same input data - I stand corrected!
So if you don't zoom one plot in but not the other, here is what you see:
2018:
2019:
These look functionally identical to me! Any minor differences between the two are because of random sampling (controlled by the --p-n parameter).
I don't think so, not in this case - the stability between the two "runs" of demux summarize support this.
Hope that helps!
2 Likes
That is the result node for the QZV itself. Click on the square above that node (like in my picture) to see the execution node for the import step. 
ah okay thanks. I always get confused with this. ha ha. Thanks so much Matthew. I really appreciate your help 
1 Like
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.