Setting up mixed DNA/Multi-State RAxML analysis

[Michael Forthman]

Hello,

I'm trying to run a ML analysis using CIPRES RAxML on a combined DNA and morphological (multi-state) dataset. I've posted to the CIPRES Google Groups, but was recommended to post here instead. I'm wanting to use a partition file (-q) and currently have it as such:

DNA, p1 = 1-1024

DNA, p2 = 1025-1855

DNA, p3 = 1856-2628

DNA, p4 = 2629-3054

MULTI, morph = 3055-3199

Here is the problem I am having trouble figuring out: I want to use a GTR+GAMMA+I for each of the four DNA partitions, but use the Lewis Mkv model (with asc correction) for the multi-state partition. Is there anyway I can specify this in the partition file? If I try selecting options in CIPRES that apply to the DNA partitions, I cannot select anything for the multi-state partitions, and vice versa. 

Cheers, Michael

[Michael Forthman]

I'm thinking it may be easier to run this on the desktop version of RAxML. However, I'm still confused as to how I can treat my partitions the way that I'm wanting to. I have to specify (-m), which only let's me specify one model. If I put GTRGAMMA, then this would be applied to each DNA and MULTI partition. But if I list MULTIGAMMA, would this apply the specified model in (-K) for MULTI partition but the GTRGAMMA model to DNA partitions (without estimating invariant sites)? 

Also, I want to include ascertainment bias for the MULTI partition. Am I correct that I simply add ASC_ before MULTI in the partition file, and then in the command line -asc_corr=lewis if this is the correction I want?

So far, this is how my command line is looking: raxmlHPC -f a -s CombinedMatrix.phy -x 2016 -p 4325 -n Result -m MULTIGAMMA -q partition.txt -asc_corr=lewis -K MK -o Peiratespunctorius -N 20

Cheers, Michael

[Alexandros Stamatakis]

On 25.01.2016 02:53, Michael Forthman wrote:
> I'm thinking it may be easier to run this on the desktop version of
> RAxML. However, I'm still confused as to how I can treat my partitions
> the way that I'm wanting to. I have to specify (-m), which only let's me
> specify one model. If I put GTRGAMMA, then this would be applied to each
> DNA and MULTI partition. But if I list MULTIGAMMA, would this apply the
> specified model in (-K) for MULTI partition but the GTRGAMMA model to
> DNA partitions (without estimating invariant sites)?

from the -m string only the model of rate heterogeneity will be
extracted, which is applied to all partitions (DNA and MULTI).

the value passed via -K then specifies which model will be applied to
the multi-state partition ...

> Also, I want to include ascertainment bias for the MULTI partition. Am I
> correct that I simply add ASC_ before MULTI in the partition file, and
> then in the command line -asc_corr=lewis if this is the correction I want?

yes, that should work.

> So far, this is how my command line is looking: raxmlHPC -f a -s
> CombinedMatrix.phy -x 2016 -p 4325 -n Result -m MULTIGAMMA -q
> partition.txt -asc_corr=lewis -K MK -o Peiratespunctorius -N 20

looks good,

alexis

>
> Cheers, Michael
>
> On Saturday, January 23, 2016 at 9:55:27 AM UTC-8, Michael Forthman wrote:
>
> Hello,
>
> I'm trying to run a ML analysis using CIPRES RAxML on a combined DNA
> and morphological (multi-state) dataset. I've posted to the CIPRES
> Google Groups, but was recommended to post here instead. I'm wanting
> to use a partition file (-q) and currently have it as such:
>
> DNA, p1 = 1-1024
> DNA, p2 = 1025-1855
> DNA, p3 = 1856-2628
> DNA, p4 = 2629-3054
> MULTI, morph = 3055-3199
>
> Here is the problem I am having trouble figuring out: I want to use
> a GTR+GAMMA+I for each of the four DNA partitions, but use the Lewis
> Mkv model (with asc correction) for the multi-state partition. Is
> there anyway I can specify this in the partition file? If I try
> selecting options in CIPRES that apply to the DNA partitions, I
> cannot select anything for the multi-state partitions, and vice versa.
>
> Cheers, Michael
>

> --
> You received this message because you are subscribed to the Google
> Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to raxml+un...@googlegroups.com
> <mailto:raxml+un...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Adjunct Professor, Dept. of Ecology and Evolutionary Biology, University
of Arizona at Tucson

www.exelixis-lab.org

[Mark Miller]

I believe it is possible to create that command line with the CIPRES interface.

Alexis, I have a follow up question.

If one wanted to run the Multi partition with an invariable option (but not the DNA partitions), could one do that?

What would the command line look like?

Mark

On Saturday, January 23, 2016 at 9:55:27 AM UTC-8, Michael Forthman wrote:

[Alexandros Stamatakis]

no one can't, the same model of rate het. will be applied to all
partitions, there is no way to specify distinct models of rate het among
distinct partitions ...

alexis

[Michael Forthman]

Thank you Alexis!

On Saturday, January 23, 2016 at 9:55:27 AM UTC-8, Michael Forthman wrote:

[Michael Forthman]

Hello again,

I have tried to start RAxML 8.1.15 on the desktop and get the following error message:

The file RAxML_flagCheck RAxML wants to open for writing or appending can not be opened [mode: wb], exiting.

This file was not included in the download. I haven't seen any posts on the matter. I tried creating a file with the file name the program is searching for, but I was still replicating the error message.

Cheers, Michael

On Saturday, January 23, 2016 at 9:55:27 AM UTC-8, Michael Forthman wrote:

[Michael Forthman]

I figured out the issue; I needed to set permission to read & write for the RAxML directory and all of it's contents.

[Alexandros Stamatakis]

correct, also please upgrade to the latest RAxML version 8.2.4:

https://github.com/stamatak/standard-RAxML/releases

alexis