Can BGM compare the specified branches (foreground) against the Background branches as done in BUSTED?

[Enhancer01]

I am running BGM in Commandline, similar to some of the models on hyphy, I saw that you can pre-select some foreground branches to test coevolution against the references ( This is actually the main reason I decided to use hyphy instead of CoMap). However, after running the program with this in mind, I discovered that there were no differences in results between specifying a branch and selecting the entire tree. My intention was to determine if coevolution is driving a particular trait in some branches compared to others.

Thank you and looking forward to your feedback.

[spond]

Dear @Enhancer01,

For BGM, specifying a set of branches does not compare them against the other branches, but rather restricts co-evolutionary analyses to substitutions occurring only on selected branches; the result should in generally differ from what happens when you run the analysis on the entire tree.

While not a proper test, you could at least explore the differences between inferred co-evolution patterns by first running the test on your desired set of branches, and then running the test on all the other branches.

Best, Sergei

[Enhancer01]

Thank you very much @spond; do you think it will be possible to incorporate this future release?

[spond]

Dear @Enhancer01,

You mean a comparison between two sets of branches? For BGM this would require a conceptually different model and it's not immediately obvious for me how to do it. We are working on some entirely novel epistatic models where you will be able to do this sort of comparison, but they probably won't be available until sometime in 2023.

I think for exploratory purposes, you could just run BGM on two sets of branches separately and see what you find. If the sets of branches are of roughly equal size (and roughly equal number of substitutions), you should get a decent idea if they are very different.

Best, Sergei

[Enhancer01]

Thank you very much, this was helpful.

[Enhancer01]

Dear @spond Please, one more thing, since I will be testing along different branches of the tree, what should be the minimum number of sequences at which the coevolution signal will be strong enough to be detected by BGM? my phylotree has a total of 20 taxa but the tested branches is made up of fewer taxa (5 or 4).

Thank you

[github-actions[bot]]

Stale issue message