Jamie Morton
This will likely mitigate some of the regularization problems that we are seeing in regression.
It would be nice to have some high level workflows in place (such as a default workflow for incorporating phylogenetic variables analysis).
Extending from @mnhuda's comment. It would be nice to have standard errors specific for the explanatory variable of interest. Not sure how this can be done - but it may...
`gneiss.util.block_diagonal` doesn't accept numpy random states. This needs to be fixed.
Looks like in one of the recent refactors, the margins of the heatmap got shifted 
If we can refactor all of the modules to have the same API as sklearn, we can use their cross-validation techniques out of the box! - [ ] couple pseudocount...
- [ ] enlargen nodes in tree - [ ] reformat html table with clear cut boundaries cc @nbokulich
Probably should add in utility method to handle tips in the `TreeNode.subset` method. https://github.com/biocore/scikit-bio/issues/1525 So it would be something like ```python def fetch_tips(t): if len(t.children) == 0: return {t.name} else:...
It would be great if we could have an automated way to subtract out residuals from submodels, so that confounders can be subtracted out from the raw data.
Need to pass in the SoftImpute module into Gneiss We'll want - [ ] Cross validation with hyperparameter tuning on the minval - [ ] A really solid model statistic...