I am new with Qiime2, but found that Qiime2 is very powerful software. I am wondering whether all the statistical analyses’ outputs from Qiime2 are publishable? Do we need to use “phyloseq” or R to do anything?
This somewhat depends on your analysis. QIIME 2 is a full pipeline: clustering/denoising and annotation, tree building, diversity calculation and visualization and several statistical tests. The plug in system means that are are more and more tools being added - useful for bringing in new techniques.
Personally, I tend to use QIIME for about 80+% of what I do, with a mix of command line and python API. I occasionally use the underlying tools because I want control of a specific aspect of the analysis that I find harder to access through the QIIME APIs.
The other place I think many people chose not to use QIIME is for plotting. Emperor works well for me, but sometimes it’s easier to adjust plots on my own in the underlying or other software than to use the ones directly from QIIME and try to adjust label, colors and fonts. Unfortunately, what looks good on my screen often isn’t readable when I go to a figure panel. So, this might be a place where you’d want to go to ggplot, or prism, or seaborn, or excel… or whatever else your favorite plotting library is, and use that.
Thank you very much for your sharing! So, If I understand well, it means that all Qiime outputs, in particular statistical analyses, can be used for publication. But if the plots don’t look good, we use other software to plot them.
Yes. Any statistical method used in QIIME 2 is a valid statistical method, published and widely used by others. So be sure to cite each individual method you use in QIIME 2, in addition to QIIME 2, to properly credit the work of others! See here for more details on discovering tools that should be cited for individual actions in QIIME 2:
Plots are highly subjective... I find most QIIME 2 plots are easily adjusted in adobe illustrator and other image editing tools to prepare publication-ready plots. But this does not cover all possible plotting modalities! So many users (myself included) will generate finished plots in R or other software.
No, you do not need to. Many of the same methods in phyloseq and other R packages are also in QIIME 2.
That said, it all depends on the needs of your analysis! QIIME 2 is in constant development and we seek to add features that are useful to microbiome scientists, but we cannot possibly contain all useful tools and statistical techniques! It is very typical for an analysis to begin in QIIME 2 and then move to R or other software for additional methods not covered in QIIME 2.