I'm trying to perform a SLAC analysis using an optimised nucleotide substitution model, following sections 2.3.7 and 2.6.2 in the "Estimating Selection Pressures" document supplied with the HyPhy installation.

Using the file InfluenzaA_H3_Random35.nex, I can create an optimised model and save it via ‘Analysis > Results >
Save Results > Export Analysis’. On running QuickSelectionDetection.bf with the parameters given in section 2.6.1, however, the analysis fails immediately after selecting 'single ancestor counting' with the message 'Can't create empty trees' (message log attached).

Looking inside the exported analysis results from the first step, I can see a tree and what appears to be the optimised model data added during the analysis, but I suspect I'm missing a step that will make this accessible to the SLAC analysis.

Would be grateful for any help

William

---

I've managed to get this to work now, by comparing the model output from nucmodelcompare.bf with a model saved from quickselectiondetection.bf.

In order to make the model from nucmodelcompare.bf acceptable to quickselectiondetection.bf, the names of three variables in the model file need to be changed:

tr to givenTree
m to NucleotideMatrix
pi to overallFrequencies

Once all references to these variables are changed in the model file, the SLAC analysis runs through without problems.

I don't know whether this is a bug or whether I am missing a step, but given that the content of the file works apparently correctly with only the variable names needing to be changed, I suspect it's a bug. At any rate maybe this info will help someone else :)

Regards

William

Hi William,

It is a bug:) We are in the middle of a massive code-base overhaul with HyPhy at the moment, so things like that do happen. Thanks for your patience and sorry for the hassle.

Sergei