Jamie Morton

I've been playing around with CTF and have occasionally noticed the following error ``` Traceback (most recent call last): File "/home/juermieboop/miniconda3/envs/qiime2-2020.8/lib/python3.6/site-packages/q2cli/commands.py", line 329, in __call__ results = action(**arguments) File "",...

Now that these functions are merged into biom-format, we can remove them from gneiss (yay).

For datasets without a clear defined phylogeny, we need to use hierarchical clustering. The scipy implementation is quite slow, but perhaps we can speed this up with fastcluster : https://pypi.org/project/fastcluster/

Renaming clades

1

Right now, the `y_i` renaming of internal nodes is a bit jarring. Perhaps we should enable renaming of only internal nodes that don't have names with something like `clade_i`. The...

There is the scenario where we will need to compute the ILR transform across an entire tensor. Pull requests will be coming up shortly.

Right now, the runtime to create a balanced tree is O(n^2). We can do better, it should be O(n).

Proportionality test

1

We will want to add some support for proportionality - [ ] Implementation of proportionality metrics - [ ] B-association regression tests

Balance test

1

One thing that could be useful is a very simple command that allows for a balance test. It would look something like this 1. Bootstrap the samples (via multinomial sampling)...

There needs to be an explicit for this. See post below for more details. https://forum.qiime2.org/t/color-correction-in-gneiss-balance-taxonomy/7218/9

numerator cutoff is slightly off in proportion_plot

2

From this forum post: https://forum.qiime2.org/t/gneiss-picking-up-of-meaningful-balances/8909/5 It looks like the numerator / denominator cutoff is slightly off -- here Bifidobacter should be in denominator - but in the plot, it looks...