MULTISTATE MODEL FOR ANALYSES OF RECODED AMINO-ACID DATA

Alexandros Vasilikopoulos

unread,

Jun 21, 2019, 8:45:57 AM6/21/19

to IQ-TREE

Dear iqtree team,

I was not able to find an appropriate answer to my question in the group so here it is: I have a long (genomic) amino-acid alignment which i would like to analyze as recoded (eg. 6 states). This means by reducing the states from 20 to 6 for example (in Dayhoff-6 recoding).

Is there a multistate model implemented in IQ-TREE that can be used to analyze such data? For example i have done such analyses in raxml before by using the -m MULTIGAMMA model and the GTR matrix. Is there a similar model implemented in IQ-TREE to do

such analyses? I guess one option is to use the morphological models available (e.g. GTR2, JC2) but those are only for 2 states (i.e. binary data). Is there a possibility to use a model that can accommodate any number of states (up to 20) depending on the analysis? Any help would be appreciated.

Many thanks

alex

Minh Bui

unread,

Jun 25, 2019, 11:36:52 AM6/25/19

to IQ-TREE, Alexandros Vasilikopoulos

Hi Alex,

Just to begin: I attended a talk at the Evolution meeting yesterday (the presenter was Alexandra Hernandez), which concluded that 6-state recoding actually increases saturation. I’m not sure how generalised this conclusion is, given their few simulations. But at least this gives us some warning, and one should compare it with the AA tree.

Now the answers:

On 21 Jun 2019, at 8:45 am, Alexandros Vasilikopoulos <alexva...@gmail.com> wrote:

Dear iqtree team,

I was not able to find an appropriate answer to my question in the group so here it is: I have a long (genomic) amino-acid alignment which i would like to analyze as recoded (eg. 6 states). This means by reducing the states from 20 to 6 for example (in Dayhoff-6 recoding).
Is there a multistate model implemented in IQ-TREE that can be used to analyze such data? For example i have done such analyses in raxml before by using the -m MULTIGAMMA model and the GTR matrix. Is there a similar model implemented in IQ-TREE to do

such analyses?

Yes there is. Once you converted the alignment into 0-5 characters, simply run IQ-TREE with "-mset GTR". By default ModelFinder only tests the MK model for morphological data. This -mset option will change this to GTR.

Cheers,

Minh

I guess one option is to use the morphological models available (e.g. GTR2, JC2) but those are only for 2 states (i.e. binary data). Is there a possibility to use a model that can accommodate any number of states (up to 20) depending on the analysis? Any help would be appreciated.

Many thanks
alex

--
You received this message because you are subscribed to the Google Groups "IQ-TREE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to iqtree+un...@googlegroups.com.
To post to this group, send email to iqt...@googlegroups.com.
Visit this group at https://groups.google.com/group/iqtree.
To view this discussion on the web visit https://groups.google.com/d/msgid/iqtree/6d9dd092-4393-4a78-8044-d2920528883d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.