Taxa abundance analysis

Hi,
I am curious about the taxa abundance analysis that reports the percentile of abundance. Is there anyway to get the statistical values or pairwise comparisons in QIIME 2?

Thanks,
BIng

1 Like

Hi @Bing! I’m not sure what analyses you’re referring to – are you asking about the QIIME 2 taxa barplots generated by qiime taxa barplot? If so, we just added an “export” feature to the taxa barplots so that you can export the underlying data that’s being plotted as a delimited text file. This export feature will be available in the next QIIME 2 release (2017.7, scheduled for sometime next week).

If that doesn’t answer your question, can you provide more details about what you’re trying to accomplish? For example, is there a corresponding QIIME 1 script you’ve used in the past, or can you point me to a paper describing or using this type of analysis?

Hello @jairideout,

I meant the last step in the tutorial: Differential abundance analysis. After the ANCOM analysis, we can find the different taxa percentile in terms of one variable such as group. My question is that whether we can get specific p-values or pairwise comparisons in the ANCOM analysis in QIIME 2 directly or we have to run it again in R?

Thanks,
Bing

Thanks for clarifying! My understanding is that ANCOM doesn’t report p-values, so I’m not sure that running ANCOM in R would get you that info.

I don’t think there’s a way to get pairwise comparisons directly in QIIME 2. The scikit-bio implementation of ANCOM (which QIIME 2 is using) doesn’t support pairwise comparisons either. I’m not certain if the R implementation supports pairwise comparisons, you may want to verify that before exporting and loading your data into R.

@mortonjt do you have any suggestions for pairwise comparisons and/or p-values in ANCOM?

@jairideout
so as presented in the tutorial example, only the significant different abundances are shown, right? If not, where can we get that “Reject null hypothesis” table based on the reference sciket-bio implementation?

Thanks,
Bing

@jairideout is right - ANCOM doesn’t support pvalues or pairwise comparisons. But you can get the W-statistic, which is a proxy for the statistical significance.

1 Like

Thank you so much! Bing

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.