Choosing appropriate substitutional models.

[Abid Ali]

Dear All, 

I hope this email finds you well.

I am currently working on molecular dating for my species using BEAST version 1.10.4. However, I am encountering challenges when attempting to apply the substitution models suggested by PartitionFinder to each partition of my dataset.

For reference, the models recommended by PartitionFinder are as follows:

#nexus

begin sets;
charset Subset1 = 1144-1803\3 1-1143\3;
charset Subset2 = 2-1143\3 1145-1803\3;
charset Subset3 = 1146-1803\3 3-1143\3;
charpartition PartitionFinder = K80+I+G:Subset1, HKY+I:Subset2, GTR+G:Subset3;
end;

I'm unsure if it's possible to incorporate all these models into the same version of BEAST (1.10.4) for my molecular dating analysis. I'm seeking advice on how to handle this situation effectively. Your valuable suggestions and guidance would be greatly appreciated!

All the best,

Abid

[Abid Ali]

Hi All!

I'm still seeking help with my question and will repost it.

I'm currently working on a molecular dating analysis using Beast 1.10.4. 

Here is the general information about the dataset;

Data= CYTB+COI Concatenated, 

bp lenght=1803, 

number of sequences= 360). 

I initially, partitioned my data into three codon positions, and, ran PartitionFinder 2.1.1 to determine the best-fit models for each. Here is the suggested model "TRNEF+I+G," "HKY+I+X," and "TRN+G+X," with respect to each partition.

I sought guidance from the Lan Fear Lab's resource (http://www.robertlanfear.com/), which provided valuable insights into mapping PartitionFinder models to BEAUTi.

According to this information, here's how the PartitionFinder models correspond to BEAUTi:

"K80 in PartitionFinder" should be represented as "HKY" in BEAUTi with "base frequencies" set to "All Equal."

"TrNef in PartitionFinder" corresponds to "TN93" in BEAUTi with "base frequencies" set to "All Equal."

"SYM in PartitionFinder" translates to "GTR" in BEAUTi with "base frequencies" set to "All Equal."

"HKY in PartitionFinder" is aligned with "HKY" in BEAUTi with "base frequencies" set to "estimated."

"TrN in PartitionFinder" corresponds to "TN93" in BEAUTi with "base frequencies" set to "estimated."

"GTR in PartitionFinder" should be represented as "GTR" in BEAUTi with "base frequencies" set to "estimated."

These guidelines helped me accurately specify the models in BEAUTi. 

However, I still have a question regarding the "+X" option, which seems necessary for partitions 2 and 3 in my analysis (i.e..., HKY+I+X," and "TRN+G+X,). 

I am not sure how to fit +X in the BEAUTI in order to fulfill the requirement for analysis.

I'm seeking suggestions on addressing this aspect, it would be greatly appreciated.

All the best,

Abid.

[Brandon Stark]

please using mega to test models.

[Dylan O'Hearn]

Hi Abid,

You might have some specific reason for wanting to use the exact models suggested by partition finder, but otherwise, the preferred method with BEAST as far as I know is to use the "bmodeltest" package, meaning you select bmodeltest as your substitution model.  This will automatically average over all available models in BEAST according to how well they fit.  You can find more information with a search on this forum or for the actual published paper on it.