Partitioned analysis with binary partition at Cipres

[die...@nocrotec.com]

Hello all,
I have a short question:
I want to configure a simple partitioned analysis at Cipres with some nucleotide partitons and one binary indel partion.
I have tried some ways, but I can't find out how to do it right.
For example on the first step I need to select the data type, but a mixed data set is not listed there.
So I thought I need to select "Nucleotide" and specify at the partition file that there is a binary partition aswell (not sure if that's right).
But then I don't know, for what partition(s) the settings in the Cipres formular belong to.
For example:
"Correct for Ascertainment bias (ASC_)": I want "yes" for the binary partition and "no" for the nucleotide partitons. But I can only set "yes" or "no".
Can someone help me how to set it up?
(Sorry for that very simple qestion - I'm new to Raxml)
All the best
Dieter

[Alexandros Stamatakis]

On 11.11.2017 21:59, die...@nocrotec.com wrote:
> Hello all,
> I have a short question:
> I want to configure a simple partitioned analysis at Cipres with some
> nucleotide partitons and one binary indel partion.
> I have tried some ways, but I can't find out how to do it right.
> For example on the first step I need to select the data type, but a
> mixed data set is not listed there.

you can select DNA and then specify a partition file that can contain
binary, DNA, and protein data ...

> So I thought I need to select "Nucleotide" and specify at the partition
> file that there is a binary partition aswell (not sure if that's right).

that's right ...

> But then I don't know, for what partition(s) the settings in the Cipres
> formular belong to.
> For example:
> "Correct for Ascertainment bias (ASC_)": I want "yes" for the binary
> partition and "no" for the nucleotide partitons. But I can only set
> "yes" or "no".

this cans also be specified in the partition file, see manual page 36:

https://sco.h-its.org/exelixis/resource/download/NewManual.pdf

alexis

> Can someone help me how to set it up?
> (Sorry for that very simple qestion - I'm new to Raxml)
> All the best
> Dieter
>

> --
> 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

www.exelixis-lab.org

[Dieter Waechter]

Hello Alexandros,
Thank you very much,
I'll try to set it up... will post here when I have been successful.
All the best
Dieter

> You received this message because you are subscribed to a topic in the
> Google Groups "raxml" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/raxml/Rk9sIQBd1Xg/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> raxml+un...@googlegroups.com.

[die...@nocrotec.com]

Hi,
I'm back. I got it work (I think). As promised, I tell my results here.
What I did:
I created a small testalignment and a partition file like that (only for testing):

DNA, partition1 = 1-21
ASC_BIN, partiton2 = 22-25

Patition1 is the standard DNA nucleotide partion, partition2 is the indel partition.

I selected RAxML-HOC2 (at Cipres)
set it up with:
Correction for ascertainment bias (ASC_): yes
Ascertainment bias correction type: Lewis (I think Lewis is good for Indel-Partition, right?)

I got:
Partition: 0
Alignment Patterns: 20
Name: partition1
DataType: DNA
Substitution Matrix: GTR

Partition: 1
Alignment Patterns: 4
Name: partiton2
DataType: BINARY/MORPHOLOGICAL
Substitution Matrix: Uncorrected
Correcting likelihood for ascertainment bias

RAxML was called as follows:
raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m ASC_GTRCAT -q part.txt -c 25 -p 54272 -f a -x 21287 --asc-corr lewis

RAxML computed a tree, I compared it with MrBayes (on which I ran a similar setup) and got 2 nearly identical trees.
So all I can ask is:
Have I forgotton anything?
Al the best
Dieter

[Alexandros Stamatakis]

> I'm back. I got it work (I think). As promised, I tell my results here.

thank you ...

> What I did:
> I created a small testalignment and a partition file like that (only for
> testing):
> |
> DNA,partition1 =1-21
> ASC_BIN,partiton2 =22-25
> |
>
> Patition1 is the standard DNA nucleotide partion, partition2 is the
> indel partition.
>
> I selected RAxML-HOC2 (at Cipres)
> set it up with:
> Correction for ascertainment bias (ASC_): yes
> Ascertainment bias correction type: Lewis (I think Lewis is good for
> Indel-Partition, right?)

If you don't know what the rest of the genome looks like, I guess so.

> I got:
> Partition: 0
> Alignment Patterns: 20
> Name: partition1
> DataType: DNA
> Substitution Matrix: GTR
>
> Partition: 1
> Alignment Patterns: 4
> Name: partiton2
> DataType: BINARY/MORPHOLOGICAL
> Substitution Matrix: Uncorrected
> Correcting likelihood for ascertainment bias
>
> RAxML was called as follows:
> raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m ASC_GTRCAT -q
> part.txt -c 25 -p 54272 -f a -x 21287 --asc-corr lewis
>
> RAxML computed a tree, I compared it with MrBayes (on which I ran a
> similar setup) and got 2 nearly identical trees.
> So all I can ask is:
> Have I forgotton anything?

Not really.

Alexis

> Al the best

[Dieter Waechter]

And again: Thank you Alexis!

All the best
Dieter


----- Original Message -----
From: "Alexandros Stamatakis" <alexandros...@gmail.com>
To: <ra...@googlegroups.com>
Sent: Sunday, December 03, 2017 11:07 PM
Subject: Re: [raxml] Partitioned analysis with binary partition at Cipres

> You received this message because you are subscribed to a topic in the
> Google Groups "raxml" group.

> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/raxml/Rk9sIQBd1Xg/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> raxml+un...@googlegroups.com.

[die...@nocrotec.com]

Hello Alexis,
I tested that now, with the following inputs and settings:

Input:
- 1 DNA partition plus one Gapmatrix (Partition 0)
- 1 partitionfile (Partition 1) reading:

DNA, LSU = 1-2331
ASC_BIN, IND_LSU = 2332-2571

Setting at Cipres:
Tool: RAxML-HPC2 on XSEDE
Data Type: Nucleotide
Use a mixed/partitioned model? (-q): The partition file above
Correct for Ascertainment bias (ASC_): yes
Estimate proportion of invariable sites (GTRGAMMA + I): no
Ascertainment bias correction type (--asc-corr): Lewis
Choose model for bootstrapping phase: GTRCAT
Conduct rapid bootstrapping? (-x): yes
Conduct a rapid Bootstrap analysis and search for the best-scoring ML tree in one single program run. (-f a): yes
Let RAxML halt bootstrapping automatically: yes
Stop Bootstrapping Automatically with Majority Rule Criterion (recommended): yes
Select Majority Rule Criterion: autoMRE

Analysis ran well, but RAxML reports:
call was:

raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m ASC_GTRCAT -q part.txt -c 25 -p 54272 -f a -x 21287 --asc-corr lewis

WARNING: you specified that you want to use an ascertainment bias correction for partition 1 for the CAT model of rate heterogeneity. Are you sure that you don't want to use a model without any rate


But I did not want and also did not specify the "CAT model" for partition 2. I want to use the 2-parameter Mkv model (Lewis 2001).

What have I done wrong?
All the best
Dieter

[Alexandros Stamatakis]

That's two different things here:

1. the substitution model (e.g. GTR, JC69, MKv)
2. the model of rate heterogeneity (CAT or GAMMA)

Alexis

> raxmlHPC-HYBRID -T 4-N autoMRE -n result -s infile.txt -m ASC_GTRCAT -q
> part.txt -c 25-p 54272-f a -x 21287--asc-corr lewis

> |
>
> WARNING: you specified that you want to use an ascertainment bias
> correction for partition 1 for the CAT model of rate heterogeneity. Are
> you sure that you don't want to use a model without any rate
>
>
> But I did not want and also did not specify the "CAT model" for
> partition 2. I want to use the 2-parameter Mkv model (Lewis 2001).
>
> What have I done wrong?
> All the best
> Dieter
>

[Dieter Waechter]

Thank you Alexis,

Yes, but I don't know how to tell RAxML (over Cipres) and the partition file what I want. I think I need to explain that more detailed:

 

What I wanted was:

 

for the DNA partition:

substitution model: GTR

Model of rate heterogeneity: CAT (or GAMMA alternatively)

ascertainment bias correction: no

--> what should be "-m GTRCAT" or alternatively "-m GTRGAMMA" at the command line.

 

for the binary partition:

substitution model: 2-paramter Mkv model

Model of rate heterogeneity: CAT (or GAMMA alternatively)

ascertainment bias correction: yes

--> what should be "-K MK --asc-corr lewis" at the command line or "ASC_BIN" at the partition file if I understood that right.

 

But the result for the command line which Cipres generated with the settings in my last post was not what I expected. What I expected was:

raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m GTRCAT -q part.txt -c 25 -p 54272 -f a -x 21287 -K MK --asc-corr lewis

 

but the command line generated at Cipres was:

raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m ASC_GTRCAT -q part.txt -c 25 -p 54272 -f a -x 21287 --asc-corr lewis

 

You see: -m ASC_GTRCAT instead of -m GTRCAT is set and the switch -K MK is missing.

 

So my questions are:

a) Is my expected command line right, or do I mix something up, and the command line Cipres generated does what I want? (I can not find that out from the log file for the substitution models for the DNA-partitions, but the line "Substitution Matrix: Uncorrected" at the Binary partitions tells me that the -K MK is missing for that partitions)

 

b) If my expected command line is right, how do I tell that Cipres and/or in the partition file?

 

Notes:

The switch "-K" is grayed out at Cipres, so I can not set it there, except if I select "Data type: Multi-State Morphological" what is also wrong.

I can set "Correct for Ascertainment bias (ASC_)" to "no" at Cipres, what would correct the -m switch to "-m GTRCAT", but if I do that, then the switch "-K MK --asc-corr lewis" is missing at the command line.

 

All the best

Dieter

 

 

----- Original Message -----

From: "Alexandros Stamatakis" <alexandros...@gmail.com>

To: <ra...@googlegroups.com>

Sent: Thursday, January 04, 2018 2:30 PM

Subject: Re: [raxml] Partitioned analysis with binary partition at Cipres

>
> --
> You received this message because you are subscribed to a topic in the Google Groups "raxml" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/raxml/Rk9sIQBd1Xg/unsubscribe.

> To unsubscribe from this group and all its topics, send an email to

raxml+un...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Mark Miller]

I want to add a piece of information: I modified the CIPRES interface so the -K option is not greyed out any longer.
This does not answer the question about how to submit when one partition requires ASC bias and one does not..
Currently the ability to select the -asccorr option requires selecting the ASC_ prefix, but perhaps I should remove that requirement as well.
But I am not sure how to handle it when one partition requires ASC and one does not.
I can adapt the interface to manage whatever you advise.

Mark

[Alexandros Stamatakis]

Dear Dieter,

Your expected command line is right, that must really be an issue with
the CIPRES interface which is beyond my control.

Alexis

On 05.01.2018 15:52, Dieter Waechter wrote:
> Thank you Alexis,
>
> Yes, but I don't know how to tell RAxML (over Cipres) and the partition
> file what I want. I think I need to explain that more detailed:
>
> What I wanted was:
>

> *for the DNA partition:*
>
> substitution model: *GTR*
>
> Model of rate heterogeneity: *CAT* (or GAMMA alternatively)
>
> ascertainment bias correction: *no*

>
> --> what should be "-m GTRCAT" or alternatively "-m GTRGAMMA" at the
> command line.
>

> *for the binary partition:*
>
> substitution model: *2-paramter Mkv model*
>
> Model of rate heterogeneity: *CAT* (or GAMMA alternatively)
>
> ascertainment bias correction: *yes*

>
> --> what should be "-K MK --asc-corr lewis" at the command line or
> "ASC_BIN" at the partition file if I understood that right.
>
> But the result for the command line which Cipres generated with the
> settings in my last post was not what I expected. What I expected was:
>
> raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m GTRCAT -q
> part.txt -c 25 -p 54272 -f a -x 21287 -K MK --asc-corr lewis
>
> but the command line generated at Cipres was:
>
> raxmlHPC-HYBRID -T 4 -N autoMRE -n result -s infile.txt -m ASC_GTRCAT -q
> part.txt -c 25 -p 54272 -f a -x 21287 --asc-corr lewis
>

> You see: -m ASC_GTRCATinstead of -m GTRCATis set and the switch -K MKis
> missing.
>
> *So my questions are:*

>
> a) Is my expected command line right, or do I mix something up, and the
> command line Cipres generated does what I want? (I can not find that out
> from the log file for the substitution models for the DNA-partitions,
> but the line "Substitution Matrix: Uncorrected" at the Binary partitions
> tells me that the -K MK is missing for that partitions)
>
> b) If my expected command line is right, how do I tell that Cipres
> and/or in the partition file?
>

> _Notes:_

>
> The switch "-K" is grayed out at Cipres, so I can not set it there,
> except if I select "Data type: Multi-State Morphological" what is also
> wrong.
>
> I can set "Correct for Ascertainment bias (ASC_)" to "no" at Cipres,
> what would correct the -m switch to "-m GTRCAT", but if I do that, then
> the switch "-K MK --asc-corr lewis" is missing at the command line.
>
> All the best
>
> Dieter
>
> ----- Original Message -----
> From: "Alexandros Stamatakis" <alexandros...@gmail.com
> <mailto:alexandros...@gmail.com>>
> To: <ra...@googlegroups.com <mailto:ra...@googlegroups.com>>
> Sent: Thursday, January 04, 2018 2:30 PM
> Subject: Re: [raxml] Partitioned analysis with binary partition at Cipres
>
> > That's two different things here:
> >
> > 1. the substitution model (e.g. GTR, JC69, MKv)
> > 2. the model of rate heterogeneity (CAT or GAMMA)
> >
> > Alexis
> >
> > On 26.12.2017 11:48, die...@nocrotec.com

> <mailto:die...@nocrotec.com>wrote:

> > www.exelixis-lab.org <http://www.exelixis-lab.org>

> >
> > --
> > You received this message because you are subscribed to a topic in
> the Google Groups "raxml" group.
> > To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/raxml/Rk9sIQBd1Xg/unsubscribe.
> > To unsubscribe from this group and all its topics, send an email to

[Mark Miller]

Thanks Alexis and Dieter, we will get this fixed in the CIPRES interface in the next day or two.

On Saturday, November 11, 2017 at 12:59:15 PM UTC-8, die...@nocrotec.com wrote:

[Dieter Waechter]

Thank you Alexis,
Anyway: we testet it - and both command lines will produce exactly the same
output (i mean 100% the same output) - but of course - it should be
corrected at Cipres to make the display correct.

> raxml+un...@googlegroups.com.

[Alexandros Stamatakis]

Thanks Dieter,

Alexis