6-state Dayhoff recoding in RaxML

[bz...@bristol.ac.uk]

Hello, 

sorry this might be a simple question and it might be possible that I simply had to  read the manual better. 

I will ask it anyway.

Is it possible to run analyses of a Dayhoff recoded dataset in Raxml?  (Using Dayhoff 6-states)?   If yes how can I do it?  

best wishes

Davide

[Alexandros Stamatakis]

Hi Davide,

Yes you can, for this you need to use a multi-state model -m MULTIGAMMA
, with encodings 0 ... 5 and tell RAxML to use GTR for this multi state
model (-K GTR).

Alexis

> --
> 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

[bz...@bristol.ac.uk]

sound good thanks for slow reply to this!

d  

[bz...@bristol.ac.uk]

Sorry I meant sorry for MY slow  reply to your answer...

[Alexandros Stamatakis]

we all need to decelerate :-)

alexis

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

[bz...@bristol.ac.uk]

Alexis, 

I have now tried this and it does work.  However, I only managed to run it assuming the alignment as one single partition.

- COMMAND:

raxmlHPC-PTHREADS -s raxml_recoded_final.phy -m MULTIGAMMA -p 12345 -n TEST -T 32

HOWEVER, for this dataset I have a Partition finder set of partitions, and as I suspect lineage-specific attractions, I was hoping to implement  the Dayhoff reconmg on the PF partitioning scheme. 

Essentially, I would like to keep the PF-defined partitions separated and apply to each a 6-state GTR matrix. Can I do that?

I tried this command but 

~/bin/raxmlHPC-PTHREADS -s raxml_recoded_final.phy -m MULTIGAMMA  -q raxml_pf_partitions.txt -p 12345 -n TEST -T 32

but I got this error:

ERROR trying to assign model 0 to position 37921 

while already model 0 has been assigned to this position

[Alexandros Stamatakis]

looks like a problem with the partition file, can you please send it to me?

alexis

On 20.01.2017 11:29, bz...@bristol.ac.uk wrote:
> Alexis,
>
> I have now tried this and it does work. However, I only managed to run
> it assuming the alignment as one single partition.
> - COMMAND:
>
> raxmlHPC-PTHREADS -s raxml_recoded_final.phy -m MULTIGAMMA -p 12345 -n
> TEST -T 32
>
>
> HOWEVER, for this dataset I have a Partition finder set of partitions,
> and as I suspect lineage-specific attractions, I was hoping to implement
> the Dayhoff reconmg on the PF partitioning scheme.
>
> Essentially, I would like to keep the PF-defined partitions separated
> and apply to each a 6-state GTR matrix. Can I do that?
>
> I tried this command but
>
> ~/bin/raxmlHPC-PTHREADS -s raxml_recoded_final.phy -m MULTIGAMMA -q
> raxml_pf_partitions.txt -p 12345 -n TEST -T 32
>
>
> but I got this error:
>
> ERROR trying to assign model 0 to position 37921
>
> while already model 0 has been assigned to this position
>
>
> On Thursday, January 19, 2017 at 9:10:17 AM UTC, Alexis wrote:
>
> we all need to decelerate :-)
>
> alexis
>

> > an email to raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml+un...@googlegroups.com <javascript:>>.

[bz...@bristol.ac.uk]

thanks!

I have sent this to you privately if you do not mind (still unpublished). 

d

[Alexandros Stamatakis]

Thanks, there is nothing apparently wrong with the partition files, so
maybe it's the alignment.

For a toy dataset like this:

10 20
Cow 013320101222010200--
Carp 013320202220023200--
Chicken 013322002202122200--
Human 013320201320323200--
Loach 013320201222020200--
Mouse 013322102220112200--
Rat 013321102220111200--
Seal 013320102222210200--
Whale 013320101220112200--
Frog 013320202220120200--

a simple partition file would look like this:

MULTI, p1=1-10
MULTI, p2=11-20

and you'd invoke raxml like this:

./raxmlHPC-AVX -m MULTIGAMMA -s multi.phy -q part -p 12345 -n Test

alexis

On 24.01.2017 11:07, bz...@bristol.ac.uk wrote:
> thanks!
> I have sent this to you privately if you do not mind (still unpublished).
> d
>
> On Monday, January 23, 2017 at 12:33:25 PM UTC, Alexis wrote:
>
> looks like a problem with the partition file, can you please send it
> to me?
>
> alexis
>

[Alexis]

hi davide,

there is a discrepancy between the number of sites in your alignment 35671 and the number of sites in your partition files, quite evidently you can't assign a site 37291 to a partition if the alignment is shorter than that ;-) 

so maybe there was a mix up in the partition file or alignment generation process ...

alexis

[bz...@bristol.ac.uk]

Thanks Alex

This is now quite confusing as I am quite certain I used that dataset...

I will re-check all my files...

d

[bz...@bristol.ac.uk]

sorted for some reason I removed constant sites when recoding. seems to be running

thanks!