-
Notifications
You must be signed in to change notification settings - Fork 0
Frequently asked questions
-
What if I have continuous sample data rather than specific groups?
-
How can I get the taxa labels of tips on either side of a node?
Please cite our paper if you use POMS:
Be sure to also cite any related tools you might have used as well, such as ape and ggtree, depending on your workflow.
Please open an issue on this repository if you would like to ask a question or make a comment.
What if I want to use a different approach for calculating (or testing for differences in) sample balances?
You can specify custom balances for all non-negligible nodes in a tree, as well as the set of balance significant nodes, and direction of the difference (i.e., group1 or group2) in the POMS_pipeline function. Check out the manual_BSNs, manual_balances, and manual_BSN_dir options in the function documentation for details. Not that this requires that node names be present in the input tree.
This means that you can use any statistical framework you want for identifying BSNs! E.g., you could perform tests that take confounding variables and repeated measures into account. Alternatively, you might want to use a tool like PhILR to compute balances using a more sophisticated approach.
You can use continuous sample data as long as this data is used to identify BSNs before running the POMS workflow. The only thing that matters is whether the BSNs can be split into one of two groups. For instance, based on a correlation test one could identify BSNs with balances that are positively or negatively associated with a given continuous sample variable. One would just need to re-code "positive" and "negative" as "group1" and "group2", and specify the BSN results with the manual_BSN_dir option (see related question above) to use this data with POMS.
You can use the node_taxa function for this purpose (see example), which is useful if you would like to see the representative taxa underlying a certain BSN, for instance.
One key way to visualize the results would be to make summaries of the number of FSNs of each type, particularly for any functions that are significant based on the multinomial test.
If you want to visualize the FSNs and BSNs for a small set of functions over a phylogenetic tree, you can see how to do that in this example.
If you have any comments or questions, please feel free to create a new issue. You can also check the list of frequently asked questions.