I have SO many issues with phyloseq as a base package but kinda love micro shades 2 level bar plots I did some reading but I may go back to numerical ecology.
Are they based on a certain 2014 paper that’s been refuted twice by chance? Also, if you don’t rarely, how do you handle the effect on depth in richness? How do you set your minimum sequencing depth for analysis or do you just combine samples w/ 100 seqs & 100K seqs?
That’s definitely where I like stacked bar plots. That was how we discovered they weren’t popping for fecal samples
1000% agree with “can’t just plot phyla”. Just not sold on krona plots
Thank you!
Thank you!
Here's what I do. I don't do stacked plots as initial look (i make krona plots for each sample). I think stacked bar plots are only useful when carefully chosen (can just plot phyla!). They are the abstract of your data, you make those when you are in the writing phase. github.com/krmaas/bioin...
random collection of scripts used to process sequences - krmaas/bioinformatics
Now my data and trash can look like a dizzying array of skittles!!!
Agree PCA is appropriate for linear data (eigenvalue based Euclidean distance ordination) not sure I see compelling arguments on NMDS (rank based scaling of arbitrary distance ordination) over PCoA (eigenvalue based ordination of arbitrary distance)
But it will saturate if my metric saturates becuse it’s rank based. So, it either arbitrarily assigns a rank, or I lose that saturation info from the shape. And my eigenvalue. And I get… computational inefficiency as a bonus?