Help with getting contrast-fel.bf to work on command line

[Emilyaoc]

Hello,

I am having trouble getting the batch file for contrast-FEL to work on the command line. I'd like to include contrast-FEL in a workflow and would rather not use the interactive menu approach in HyPhy (though I know this is possible). I'd like to include it as a stand alone analysis. The line of code I have been trying is:

'hyphy contrast-fel.bf --alignment my_alignment.fas --tree my_tree.nwk --branch-set my_label'

I downloaded the batch file for contrast-FEL (contrast-fel.bf) from hyphy-analyses/SimulateMG94/modules/site-variation/contrast-fel/

Are you able to help me understand why this isn't working? I used a similar approach when using other stand alone analyses such as BUSTED-PH.bf & label-tree.bf and they worked fine.

Thanks for any help you can offer

Emily

[spond]

Dear @Emilyaoc,

contrast-fel is a standard analysis. You can run it like so

hyphy contrast-fel --alignment my_alignment.fas --tree my_tree.nwk --branch-set my_label

The actual batch file is in (if you did a standard install) /usr/local/share/hyphy/TemplateBatchFiles/SelectionAnalyses/contrast-fel.bf, but HyPhy will recognize analysis names to save you the trouble of locating the actual file.

If you type hyphy --help you will also see a long list of standard analyses you can access this way.

Best,
Sergei

[Emilyaoc]

Dear Sergei,
Ahh, I see. Sorry I misunderstood that. Another thing I am a little unsure of then is how to specify more than one set of test branches on the command line as --branch-set doesn't seem to be able to accept more than one value? I tried '--branch-set label1 label2' but that didn't work. I guess I specified the different labels (label1 and label2) incorrectly somehow?
Thank you for your help
Emily

[spond]

Dear @Emilyaoc,

For analyses that support multiple partitions you specify it like so

--branch-set label1 --branch-set label2 ...

Best,
Sergei

[Emilyaoc]

Dear Sergei,

Ok, thank you! In the scenario that your tip labels fall into one of two categories, would it then be appropriate to specify just one set of test braches to compare the two categories? Or two sets of test branches (i.e. each of the two categories)? I'm not sure whether the internal braches need to be considered in this scenario?

Thank you

Emily

[spond]

Dear @Emilyaoc,

Two sets of branches. contrast-fel will not consider "nuisance" branches for testing, so whatever you leave unlabeled should not contribute to the analysis.

Best,
Sergei

[Emilyaoc]

Great! Thank you for your help.