Hi, I am learning how to use Gneiss in Python. I was wondering if there is a way to add labels to the radialplot so that instead of hovering to see the balances, these would be displayed on the plot.
Also, how should this plot be incorporated into a publication? (This follows from my previous question.) I imagine it would be easier for a reader to interpret the plot if the balances were displayed on the image.
Short answer - yes it is possible. Long answer, the dendrogram plotting functions at the moment cannot perform those sorts of plotting functionalities
right now, the gneiss has its own Dendrogram object that can calculate the coordinates for all of the nodes,
so it should be straightforward to take the coordinates from this function, and plug it into another rendering engine instead of Bokeh
such as matplotlib
but we don't support publication ready figures with phylogenetic trees at the moment just yet -- that is something we hope to knock out in the next few upcoming months
but, if you need something to do labeling for relatively small trees, might be worth checking out ete3 or ggtree. Both of them are quite comprehensive when it comes to generating publication quality trees
Hi Jamie. Thank you for your responses. I may want to use a tree graphic from Gneiss when it is available, so if you can keep us updated, that would be great!
I have another question. In the paper, "Discrete False-Discovery Rate Improves Identification of Differentially Abundant Microbes," it is mentioned that this FDR method can be incorporated into Gneiss. Is there a tutorial out there on how to do this? I noticed that I have hundreds of numerator taxa in the first balance in my data, and I am hoping to narrow that number down, somewhat, perhaps through the use of an FDR procedure.
right - the FDR approach can accept in a table as input
it is possible to unpack the balances.qza artifact and feed that in directly
a similar approach has been hinted at in the qiime2 forums
for additional help on how to use dsfdr, we can answer questions under the issue tracker here
so that we don't get the issues mixed up
Hi guys, thank you for all the exceptional work. I was wondering: wouldn't it be great to have the Adjusted R-square in the summary? Having the r-squared only is not helpful, if I have dozens of parameters (i.e. ions, metabolites and treatments) in my model, it would be good to know which one should be removed from the model. Or should it be done balance by balance? I am writing a jupyter to do it manually, but I was wondering if you thought of it as well.
Right - need to be careful with R^2 just for that reason. But it still can be useful since it represents the variance explained by the model
we are definitely open for to add additional test statistics -- pull requests are welcome