Mirror-image reversal of ordination data and PCoA graphs from core metrics

[MicrobeManager]

Hi Justine,

Thanks for the very detailed reply, this makes more sense now. I did re-rarify each group of samples using only core-metrics. In fact, I created new manifests and did everything from the ground up for each combination of treatments I wanted to analyze. I am still new to QIIME2 so I wasn't sure how to break a larger dataset into smaller groups and pick and choose which to analyze for the core-metrics step. I probably did it the long way but I don't know all the ins and outs yet.

@jwdebelius

Now, you've probably changed the distance matrix, which changed the ordination, which can lead to flipping in the ordination.

I don't think I fully understand this though. The file manifest was the exact same as the one that produced the mirrored samples, and the results from DADA2 and everything else were the same. How would this lead to the ordination being flipped back when I reloaded it?

@jwdebelius

And then, remember you can always flip the visualization in the emperor plot. Go to the "axes" tab and there's a checkbox that lets you flip things. There are lots of reasons you might want to change the direction of an axis and it's pretty easy. If you're working programatically, the axes can be multiplied by -1.

This is good to know. I used code from this post that extracts the PC1 and PC2 values from the .qza and am graphing 2D PCoAs in R for publication. So I would just multiply the PC1 by -1 to flip them back?

Thanks again for your help!