Query about the I*R rate-heterogeneity across sites model

[lsjermiin]

Hi,

I am using an alignment of amino acids to test the performance of ModelFinder (implemented in IQ-TREE multicore version 2.3.4 COVID-edition for Linux x86 64-bit built Apr 26 2024).

In the present case, I used the following command:

iqtree2 -s Aligned_proteins_concatenated.fst -st AA -m MF -mset mtVer -mrate *G,I*G,*R,I*R,*H -mfreq FU,FO,F -mtree -T 16 -merit AIC -safe

The program aborted while processing the mtVer+FO+I*R2 model (alignment and log file attached).

Is this because the I*R model has not been implemented or is there a bug in the code?

Incidentally, the program completed its task under the BIC criterion.

Feedback would be most welcome.

Lars

05_MF3_RHAS_AIC.zip

[bqminh]

Sorry for the delay. From the log, it looks like +FO does not work with *G or *H. The likelihood doesn't make sense as it decreases during the optimisation. So please refrain from this combination of model.