RAxML-NG has three versions of the TIM model available, with equal and unequal frequencies for each (the chart linked in their manual is below). Does one of these TIM models correspond to the TIM model in PartitionFinder?
Thanks for the help!
Model | Reference | d.f. | Base freq | Symmetries | code
TIM1 | Posada, 2003 | 3 | equal | AC=GT ; AT=CG ; AG ; CT | 012230
TIM1uf | Posada, 2003 | 6 | unequal | AC=GT ; AT=CG ; AG ; CT | 012230
TIM2 | | 3 | equal | AC=AT ; CG=GT ; AG ; CT | 010232
TIM2uf | | 6 | unequal | AC=AT ; CG=GT ; AG ; CT | 010232
TIM3 | | 3 | equal | AC=CG ; AT=GT ; AG ; CT | 012032
TIM3uf | | 6 | unequal | AC=CG ; AT=GT ; AG ; CT | 012032
Hi There,
If you’re using the TIM model in PartitionFinder then you’ll be using the PhyML implementations.
These correspond to TIM1 in what you posted above (code 012230), and the base frequencies can be equal (TIMEF) or not (TIM).
If you’re ever confused about exactly what happens in PartitionFinder when you choose a model, then you can take a look at the models.csv
file which is in the /partfinder
folder and also here: https://github.com/brettc/partitionfinder/blob/master/partfinder/models.csv. It will show you for each model, rate distribution parameters, and other settings, exactly what commandline PartitionFinder uses in PhyML and (if applicable) RAxML.
Rob