PF2 suggested models for ML

[Abid Ali]

Hi there, 

I am currently working on a phylogenetic analysis using two genes, COI and CYTB (concatenated). I used Partitionfinder 2.1.1 to determine the best models for each gene's 1st, 2nd, and 3rd codon positions. I have attached the configuration input file (cfg) I used, as well as the output best scheme file for reference.

After fitting the model=All, I obtained the following suggested models for IQtree:

#nexus
begin sets;
charset Subset1 = 1144-1803\3 1-1143\3;
charset Subset2 = 1145-1803\3 2-1143\3;
charset Subset3 = 1146-1803\3 3-1143\3;
charpartition PartitionFinder = TRNEF+I+G:Subset1, HKY+I:Subset2, TRN+G:Subset3;
end;                                                                                 

However, the analysis is taking an unusually long time, nearly more than ten days, and it has not yet completed 5000 bootstraps despite utilizing more than 50 CPUs. 

This prolonged runtime seems unusual, and I suspect that something related to the partitions might be causing this delay---??.

Additionally, I performed another analysis by fitting the model=beast to choose the best model for Beast/BEAUti. The suggested models were as follows:

TRNEF+I+G
HKY+I+X
TRN+G+X

During my research, I came across information from The Lan Fear Lab (http://www.robertlanfear.com/) that provided mappings between models from PartitionFinder and BEAUti. This information helped me set the models in BEAUti correctly based on the PartitionFinder results.

However, I encountered a challenge in setting the +X option in BEAUti, which is required for partitions 2 and 3 in my case.

Therefore, my second question is: How can I set the +X option in BEAUti?

I would greatly appreciate your guidance and advice on these issues. Your suggestions will be invaluable in resolving these challenges.

Thank you very much for your time and assistance!

Best regards,

Abid

[Rob Lanfear]

Hi Abid,

Thanks for the email. Unfortunately I have no idea about Beauti - it's been a very long time since I used it. That sounds like a question for the BEAST user group.

As for the IQ-TREE question, how long your analysis takes depends on many things, including the number of taxa, the model, and the number of bootstraps. I suspect that you have a lot of taxa? Either way, 5000 bootstraps is an enormous number, and unless you want to measure bootstrap proportions to the nearest 1/5000, it's not necessary. Usually 100 would be enough. 

If this doesn't help, then the best thing is to post on the IQ-TREE user group for that question.

Rob