Hello again
I am here with a new question about the interpretation of qiime2 results
I wanna determine the count (the number) of common features (but not OTUs essentially) between the samples and then put the counts in statistical analysis like ANOVA and Doncken test to find the relationship between my samples according to the number of common features exist in the samples…
Would you please let me know which output file should i use to interpret this idea from?
Is there a ANOVA test accessible through Qiime2commands?
My next question is about sequence quality. I received my Miseq files as separated demultiplexed files from the company and after making its manifest file i just made demux summarized file and then used dada2 to trim the sequences. please let me know is there any other quality check to be done necessarily?
How may i filter low abundant features and which parameters should i use to do that?
I almost forgot to ask about the Q20/Q30 quality i was given by the company. Does mentioning those Q 20/30 quality scores mean that the low quality reads are already removed or I must remove them by Qiime?
There is not a method to find shared features yet, but there is an open issue for this. In the mean time, you can calculate Jaccard distance between each pair of samples using q2-diversity — this is effectively the ratio of features (not necessarily OTUs) that are shared between samples. You can then use diversity beta-group-significance to test whether Jaccard distance is significantly different between/within metadata groups or beta mantel to determine whether Jaccard distance correlates with another distance metric (including with continuous metadata categories).
You can read descriptions of all currently available plugins here — you can use that page as a resource for exploring other functions in QIIME 2. And of course, the tutorials describe many of these in detail, including the actions that I mentioned above.
No, not directly. Some QIIME 2 visualizers incorporate ANOVA tests, but none to perform what I describe above — the reason being that ANOVA is not an appropriate test for comparing pairwise distances. Use PERMANOVA via beta-group-significance.
If you use dada2, other quality control steps are theoretically unnecessary. dada2 detects erroneous reads and chimera — but of course other steps, e.g., removing contaminant reads, may be needed if you detect a problem.
No, most likely not. You should still perform quality control steps in QIIME 2: either q2-quality-filter or q2-dada2 will take care of what you need; follow the tutorials to learn more about using these plugins with your data.
In the future please separate each question into a separate thread; this will make it easier for other users with similar questions to follow in your footsteps in the future. Thank you for your help.
Hi Nicholas_Bokulich
I am very thankful to you for detailed explanation of the issue. That was great help for me. Of course i will send my question in separate parts.
Hi Nicholas_Bokulich
Would you please let me know what exactly is the skbio.diversity.beta_diversity command at the open issue page(for share features) you mentioned in replying to my question? May i just rely on Jaccard distance to find shared features between samples?
Thank you in advance for your great comments
Sajjad
Hi @sajjad.sarikhan, skbio.diversity.beta_diversity is the function that is actually performing the beta diversity calculation in q2-diversity — the issue page that I shared describes how this solution might be implemented by developers but is not really a workaround for qiime2 users.
Yes, and this is already accessible through the diversity beta action. This will give you the proportion of features that are shared between samples, though, not the count (or the names of features that are shared). When other methods are implemented that answer those questions, we will let you know on this thread.
