Morphological characters with ordered and un-ordered traits

[Matthew Forister]

I am just looking into IQ-TREE for the first time, and I'm hoping that someone could save me a ton of time by answering what is hopefully an easy question and maybe even supplying an example input file...

Question: is it possible to analyze an input file of morphological characters in which some of the characters are ordered (1 goes to 2 goes to 3, etc.) and some of the characters are not ordered (1 can change to 2 or 3, etc.)?

If that is possible, then it would be wonderful if someone had an example datafile with this kind of data.  Or even just a datafile with morphological characters (regardless of ordered or un-ordered status).

thank you!

Matt

[Bui Quang Minh]

Dear Matt,

1) The morphological alignment can be in Phylip, Fasta, Nexus and other formats supported by iqtree. For example:

Phylip:

4 10

S1   0123401234

S2   0324120432

S3   3202304130

S4   4230120340

2) Mixing ordered and unordered characters are currently not possible. Actually IQ-TREE should allow users to input an arbitrary rate matrix specified from a file (just like for DNA and protein model). I don’t know why this is not supported. Will let the developer know about this ;-)

M

--
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.
For more options, visit https://groups.google.com/d/optout.

--

Bui Quang Minh

Center for Integrative Bioinformatics Vienna (CIBIV)

Campus Vienna Biocenter 5, VBC5, Ebene 1

A-1030 Vienna, Austria

Phone: ++43 1 4277 74326

Email: minh.bui (AT) univie.ac.at

[Matthew Forister]

Thanks for your quick reply!  And that's too bad about mixing ordered and unordered traits.  I wonder if anyone knows of another software that could handle those data types for morphological analysis?

[Bui Quang Minh]

Hi Matt,

Can you give more details, like how the rate matrix Q looks like for such data? As said, we can of course support arbitrary rate matrix.

Minh

[Matthew Forister]

Hi Minh,

I'm afraid I do not have an outline of a rate matrix.  For now, I only have the need to have ordered and un-ordered characters.  But thank you for your responses,

Matt

[ear...@gmail.com]

Hi Minn and Matt,

MrBayes can do this, though I haven't tried it and I don't know about the details of the implementation.

Would it make sense to have two matrices sharing a parameter? For instance:

Ordered:
* a 0 0 0
a * a 0 0
0 a * a 0
0 0 a * a
0 0 0 a *
 

Unordered:

* a a a a 
a * a a a 
a a * a a 
a a a * a 
a a a a * 

Maybe in that case an asymmetric ordered model could be implemented without the need of an extra rate parameter by fixing one rate to 1:

Ordered, asymmetric:

* a 0 0 0 
1 * a 0 0 
0 1 * a 0 
0 0 1 * a 
0 0 0 1 *

Cheers,

Eduardo

[Bui Quang Minh]

Hi Eduardo,

The first two matrices you listed (ordered and unordered) are exactly those implemented in IQ-TREE (ORDERED and MK model). With a note that entry a is rescaled such that branch lengths are interpreted as the number of substitution per site. So in fact these 2 matrices have 0 free parameter. 

I don’t know about the last one (asymmetric) but it looks like a non-reversible model.

I think a mixed rate matrix may look like this:

* a 0 0 0 
a * a 0 0 

0 a * a a 

0 0 a * a 

0 0 a a * 
 

(i.e, states 1,2,3 are ordered, whereas 3,4,5 are unordered).

Minh

[ear...@gmail.com]

Hi Minh,

Thanks for your reply.

I think a mixed rate matrix may look like this:
* a 0 0 0 
a * a 0 0 
0 a * a a 
0 0 a * a 
0 0 a a * 

That makes sense! I was thinking of trying to use the two matrices simultaneously, linked by the parameter a... if that would make any sense. In any case, it would be more complicated than necessary, given your mixed rate matrix example. It would be great to have the ability to use user-supplied matrices like that.

I don’t know about the last one (asymmetric) but it looks like a non-reversible model.

Yes, sorry about that. I was thinking about a problem with rooted trees before I wrote that and forgot that the models must be reversible.

Cheers,

Eduardo