Marginal ancestral reconstruction from coding sequences

Adrian Baez-Ortega

unread,

Sep 20, 2016, 8:54:31 AM9/20/16

to raxml

Hello,

I am trying to adapt a pipeline that uses RAxML to generate a ML tree from a nucleotide alignment, and then uses PAML to both adjust the branch lengths and compute ancestral sequences in that tree. The problem is that PAML cannot scale up to the amount of data that I'm using, so I've turned to RAxML's marginal ancestral state reconstruction to save the day. I think it works very well and gives a neater output than PAML, but I have a doubt about the model.

I have been using as input sequences the concatenation of many CDS sequences with the present substitutions imposed. The reason for this is that PAML can be set up to treat a nucleotide alignment as a coding sequence alignment, as described in the PAML manual (page 25).

For nucleotide based (baseml) analysis of protein coding DNA sequences (option GC in the sequence data file), the program also calculates the posterior probabilities of ancestral amino acids. In this analysis, branch lengths and other parameters are estimated under a nucleotide substitution model, but the reconstructed nucleotide triplets are treated as a codon to infer the most likely amino acid encoded. Posterior probabilities for stop codons are small and reset to zero to scale the posterior probabilities for amino acids. A nucleotide substitution model that is very close to a codon-substitution model can be specified as follows. You add the option characters GC at the end of the first line in the data file and choose model = 4 (HKY85) and Mgene = 4. The model then assumes different substitution rates, different base frequencies, and different transition/transversion rate ratio (kappa) for the three codon positions.

I am fearing that I will lose this codon-substitution-like model by switching to RAxML. I guess that (although I'm using GTRGAMMA to build the tree and also perform marginal reconstruction) I might be able to do something to get close to the PAML approach. It would be necessary to somehow specify the HKY85 model and all those different rates/frequences/ratios for the three codon positions, but I don't know the best way to do it. On the other hand, perhaps you think that this is not necessary or that the GTRGAMMA model will behave well enough for this kind of data. I think in this case it is important to treat the sequence as a coding sequence, because I want to identify amino acid changes in the branches of the tree, not just nucleotide changes.

So to sum up, do you think this is possible/necessary with RAxML, and how would you do it? Also, do you think there is a way to better estimate the branch lengths based on this process, as PAML does?

Thanks a lot for your help!

Best,
Adrian

Alexey Kozlov

unread,

Sep 21, 2016, 4:19:18 PM9/21/16

to ra...@googlegroups.com

Dear Adrian,

I'm afraid RAxML cannot give you probabilities of ancestral aminoacids based on nucleotide alignment.

However, you can assign different models to 1st, 2nd and 3rd codon positions and thus have independent estimates for
substitution rates and other model parameters. You can do it by using a partition file (-q option) like this:

DNA, p1=1-100\3
DNA, p2=2-100\3
DNA, p3=3-100\3

You can also enforce HKY85 model for all partitions by using "--HKY85" switch (default for DNA is GTR).

Hope this answers your question at least partially, and please refer to the RAxML manual for details.

Best,
Alexey

On 20.09.2016 14:54, Adrian Baez-Ortega wrote:
> Hello,
>
> I am trying to adapt a pipeline that uses RAxML to generate a ML tree from a nucleotide alignment, and then uses PAML to
> both adjust the branch lengths and compute ancestral sequences in that tree. The problem is that PAML cannot scale up to
> the amount of data that I'm using, so I've turned to RAxML's marginal ancestral state reconstruction to save the day. I
> think it works very well and gives a neater output than PAML, but I have a doubt about the model.
>
> I have been using as input sequences the concatenation of many CDS sequences with the present substitutions imposed. The
> reason for this is that PAML can be set up to treat a nucleotide alignment as a coding sequence alignment, as described
> in the PAML manual (page 25).
>
> For nucleotide based (baseml) analysis of protein coding DNA sequences (option GC in the sequence data file), the
> program also calculates the posterior probabilities of ancestral amino acids. In this analysis, branch lengths and
> other parameters are estimated under a nucleotide substitution model, but the reconstructed nucleotide triplets are
> treated as a codon to infer the most likely amino acid encoded. Posterior probabilities for stop codons are small

> and reset to zero to scale the posterior probabilities for amino acids. *A nucleotide substitution model that is
> very close to a codon-substitution model can be specified as follows*. You add the option characters GC at the end
> of the first line in the data file and choose model = 4 (*HKY85*) and Mgene = 4. *The model then assumes different

> substitution rates, different base frequencies, and different transition/transversion rate ratio (kappa) for the

> three codon positions*.

>
>
> I am fearing that I will lose this codon-substitution-like model by switching to RAxML. I guess that (although I'm using
> GTRGAMMA to build the tree and also perform marginal reconstruction) I might be able to do something to get close to the
> PAML approach. It would be necessary to somehow specify the HKY85 model and all those different rates/frequences/ratios
> for the three codon positions, but I don't know the best way to do it. On the other hand, perhaps you think that this is
> not necessary or that the GTRGAMMA model will behave well enough for this kind of data. I think in this case it is
> important to treat the sequence as a coding sequence, because I want to identify amino acid changes in the branches of
> the tree, not just nucleotide changes.
>
> So to sum up, do you think this is possible/necessary with RAxML, and how would you do it? Also, do you think there is a
> way to better estimate the branch lengths based on this process, as PAML does?
>
> Thanks a lot for your help!
>
> Best,
> Adrian
>

> --
> 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.

Adrián Báez Ortega

unread,

Sep 22, 2016, 8:27:12 AM9/22/16

to ra...@googlegroups.com

Dear Alexey,

Thank you very much for your answer, I think that will work. But do you agree that this is a good thing to do when the input is a coding sequence alignment? I think I'll try both ways (GTRGAMMA and HKY98 and partitions) and see if there's any difference.

Thanks!

Best,

Adrian

To unsubscribe from this group and stop receiving emails from it, send an email to raxml+unsubscribe@googlegroups.com
<mailto:raxml+unsubscribe@googlegroups.com>.

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

--
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/qXTeKk7ppGM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+unsubscribe@googlegroups.com.

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

--

Adrian Baez-Ortega
linkedin.com/in/baezortega

www.divulgatum.com

Alexey Kozlov

unread,

Sep 22, 2016, 8:53:58 AM9/22/16

to ra...@googlegroups.com

Dear Adrian,

> Thank you very much for your answer, I think that will work. But do you agree that this is a good thing to do when the
> input is a coding sequence alignment?

If you mean partitioning the alignment by codon position - then yes, it's a common practice, since evolution speed and
other parameters may differ substantially (e.g. 3rd position usually being much more variable). The only caveat is that
you have 3x more parameters to estimate, but if your alignment is not very short, it shouldn't be a problem. In fact, if
you have enough data, you can even try unlinked branch length mode (-M option), which will give you independent branch
length estimates for each codon position.

>I think I'll try both ways (GTRGAMMA and HKY98 and partitions) and see if there's
> any difference.

Yes, that's always the better way :)

Best,
Alexey

> raxml+un...@googlegroups.com <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com <mailto:raxml%2Bunsu...@googlegroups.com>>.
> For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.

>
>
> --
> 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/qXTeKk7ppGM/unsubscribe

> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
> To unsubscribe from this group and all its topics, send an email to raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
>
>
>
>
> --
> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> www.divulgatum.com <http://www.divulgatum.com>

>
> --
> 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>.

Alexandros Stamatakis

unread,

Sep 22, 2016, 9:42:25 AM9/22/16

to ra...@googlegroups.com

On 22.09.2016 14:53, Alexey Kozlov wrote:
> Dear Adrian,
>
>> Thank you very much for your answer, I think that will work. But do
>> you agree that this is a good thing to do when the
>> input is a coding sequence alignment?
>
> If you mean partitioning the alignment by codon position - then yes,
> it's a common practice, since evolution speed and other parameters may
> differ substantially (e.g. 3rd position usually being much more
> variable). The only caveat is that you have 3x more parameters to
> estimate, but if your alignment is not very short, it shouldn't be a
> problem. In fact, if you have enough data, you can even try unlinked
> branch length mode (-M option), which will give you independent branch
> length estimates for each codon position.
>
>> I think I'll try both ways (GTRGAMMA and HKY98 and partitions) and see
>> if there's
>> any difference.
>
> Yes, that's always the better way :)

exactly :-) and if you do so, please post what you found here as well,

thanks,

alexis

--
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

Adrian Baez-Ortega

unread,

Sep 23, 2016, 5:57:45 AM9/23/16

to raxml

Dear Alexey and Alexis,

Thanks for all the help! I think I have enough data to use the -M option. The doubt I have is whether -M can be used with -f A, because -f A doesn't give you branch lengths, at least by default.

What I am currently doing with this coding sequence alignment is:

raxml-HPC-SSE3 -f a -m GTRGAMMA -# autoMRE to obtain the ML tree.
raxml-HPC-SSE3 -f A -m GTRGAMMA to do ancestral state reconstruction on the ML tree.

So I guess you mean using -M in the first step? My idea was replacing GTRGAMMA by HKY85 and doing the partitioning for the second step, but I don't know if it would be convenient doing it also for the tree search. The pipeline I'm using as a reference used GTRGAMMA for the tree search and then PAML with the described settings (HKY85) for the sequence reconstruction and branch length estimation. It would be good if I could refine the branch lengths either by adding the -M option in -f A or in -f a or by other means, but I would need to know whether I should also include HKY85 and partitioning in the first step. (The tree that I'm obtaining with GTRGAMMA makes a lot of sense, though.)

I hope I've explained the problem well enough...

Thank you,
Adrian

>> <mailto:raxml%2Bu...@googlegroups.com>
>> <mailto:raxml+un...@googlegroups.com
>> <mailto:raxml%2Bu...@googlegroups.com>>.

>> For more options, visit https://groups.google.com/d/optout
>> <https://groups.google.com/d/optout>.
>>
>>
>> --
>> 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/qXTeKk7ppGM/unsubscribe
>> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
>> To unsubscribe from this group and all its topics, send an email
>> to raxml+un...@googlegroups.com

>> <mailto:raxml%2Bu...@googlegroups.com>.

Alexey Kozlov

unread,

Sep 23, 2016, 6:25:48 AM9/23/16

to ra...@googlegroups.com

Hi Adrian,

yes your explanation is very clear, thanks!

I just check the code and as far as I can see "-f A" option can be combined with "-M". Moreover, "-f A" will re-estimate
branch lengths and model parameters from scratch, i.e. the only thing which is transferred from your step #1 is the tree
topology.

Which means that if you're happy with the topology you've got with unpartitioned alignment under GTRGAMMA, you can just
use it for the step #2, with the command line like:

*raxml-HPC-SSE3 -f A -m GTRGAMMA --HKY85 -M*

But you can also experiment a bit and check if tree inference under partitioned model with HKY85 and "-M" would yield a
different topology in step #1 ...

Best,
Alexey

On 23.09.2016 11:57, Adrian Baez-Ortega wrote:
> Dear Alexey and Alexis,
>

> Thanks for all the help! I think I have enough data to use the *-M* option. The doubt I have is whether *-M* can be used
> with *-f A*, because *-f A* doesn't give you branch lengths, at least by default.

>
> What I am currently doing with this coding sequence alignment is:
>

> *raxml-HPC-SSE3 -f a -m GTRGAMMA -# autoMRE *to obtain the ML tree.
> *raxml-HPC-SSE3 -f A -m GTRGAMMA* to do ancestral state reconstruction on the ML tree.
>
> So I guess you mean using *-M* in the first step? My idea was replacing GTRGAMMA by HKY85 and doing the partitioning for

> the second step, but I don't know if it would be convenient doing it also for the tree search. The pipeline I'm using as
> a reference used GTRGAMMA for the tree search and then PAML with the described settings (HKY85) for the sequence
> reconstruction and branch length estimation. It would be good if I could refine the branch lengths either by adding the

> *-M* option in *-f A* or in *-f a* or by other means, but I would need to know whether I should also include HKY85 and

> >> On 21 September 2016 at 21:19, Alexey Kozlov <alexei...@gmail.com <javascript:>

> >> raxml+un...@googlegroups.com <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com <javascript:>>
> >> <mailto:raxml+un...@googlegroups.com <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com <javascript:>>>.

> >> For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
> >> <https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
> >>
> >>
> >> --
> >> 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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>
> >> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>.
> >> To unsubscribe from this group and all its topics, send an email

> >> to raxml+un...@googlegroups.com <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com <javascript:>>.

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

> >> <https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
> >>
> >>
> >>
> >>
> >> --
> >> Adrian Baez-Ortega

> >> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>
> >> www.divulgatum.com <http://www.divulgatum.com> <http://www.divulgatum.com>

> >>
> >> --
> >> 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 <javascript:>
> >> <mailto:raxml+un...@googlegroups.com <javascript:>>.

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

Adrián Báez Ortega

unread,

Sep 23, 2016, 6:40:32 AM9/23/16

to ra...@googlegroups.com

Hi Alexey,

Many thanks for replying so quickly!

I've been so naive as to forget to mention the most important thing about my data: it's a coding sequence containing only somatic mutations from cancer samples (it's a very old contagious cancer lineage with an actual phylogeny), so probably the partitioning doesn't apply in this context, because negative selection is extremely weak or nonexistent in cancer, so we don't expect different mutation rates for each codon position. My mistake was trying to exactly replicate another pipeline that was conceived for flu data!

What I don't know if I should still keep the HKY85 model, after all. I don't know if it depends more on selection or just on the fact that it is a coding sequence.

Best,

Adrian

To unsubscribe from this group and stop receiving emails from it, send an email to raxml+unsubscribe@googlegroups.com
<mailto:raxml+unsubscribe@googlegroups.com>.

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

--

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/qXTeKk7ppGM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+unsubscribe@googlegroups.com.

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

Alexandros Stamatakis

unread,

Sep 23, 2016, 7:29:55 AM9/23/16

to ra...@googlegroups.com

okay, so in that case what you might want to take into account is
to correct for ascertainment bias, see here:

http://sysbio.oxfordjournals.org/content/64/6/1032

also our experience with this kind of SNP datasets is that

1. you probably don't need a model of rate heterogeneity
2. simpler models like HKY85 or even JC might be more appropriate

what you should really do is to run Modeltest prior to the actual tree
inference,

alexis

On 23.09.2016 12:40, Adrián Báez Ortega wrote:
> Hi Alexey,
>
> Many thanks for replying so quickly!
>
> I've been so naive as to forget to mention the most important thing
> about my data: it's a coding sequence containing only somatic mutations
> from cancer samples (it's a very old contagious cancer lineage with an
> actual phylogeny), so probably the partitioning doesn't apply in this
> context, because negative selection is extremely weak or nonexistent in
> cancer, so we don't expect different mutation rates for each codon
> position. My mistake was trying to exactly replicate another pipeline
> that was conceived for flu data!
>
> What I don't know if I should still keep the HKY85 model, after all. I
> don't know if it depends more on selection or just on the fact that it
> is a coding sequence.
>
> Best,
> Adrian
>
>
> On 23 September 2016 at 11:25, Alexey Kozlov <alexei...@gmail.com

> <alexei...@gmail.com <mailto:alexei...@gmail.com> <javascript:>
>
> >> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>

> <mailto:raxml%2Bun...@googlegroups.com> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com> <javascript:>>
> >> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bun...@googlegroups.com> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com> <javascript:>>>.

> <mailto:raxml%2Bun...@googlegroups.com> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com> <javascript:>>.

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

> >> <https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>>>.
> >>
> >>
> >>
> >>
> >> --
> >> Adrian Baez-Ortega
> >> linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>>
> >> www.divulgatum.com <http://www.divulgatum.com>
> <http://www.divulgatum.com> <http://www.divulgatum.com>
> >>
> >> --
> >> 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%2Bun...@googlegroups.com> <javascript:>
> >> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bun...@googlegroups.com> <javascript:>>.

> >> For more options, visit
> https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <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>
> <http://www.exelixis-lab.org>
>
> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com

> <mailto:raxml%2Bunsu...@googlegroups.com>>.

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

> 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/qXTeKk7ppGM/unsubscribe

> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
> To unsubscribe from this group and all its topics, send an email to
> raxml+un...@googlegroups.com

> <mailto:raxml%2Bunsu...@googlegroups.com>.

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

> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>

> www.divulgatum.com <http://www.divulgatum.com>

>
> --
> 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.

--

Adrián Báez Ortega

unread,

Sep 23, 2016, 7:35:33 AM9/23/16

to ra...@googlegroups.com

Hi Alexis,

Sorry, what I mean is not that the sequences only consist of mutated bases, but that they are coding sequences (concatenations of many CDS sequences) where the only mutations present are somatic mutations (so, no germline). We discussed this kind of data some weeks ago and you concluded that it's better not to use asc. correction bias if I have the full sequences.

Many thanks,

Adrian

it, send an email to raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>
<mailto:raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>>.

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

--
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/qXTeKk7ppGM/unsubscribe
<https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
To unsubscribe from this group and all its topics, send an email to

raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>.

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

--
Adrian Baez-Ortega
linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
www.divulgatum.com <http://www.divulgatum.com>

--
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+unsubscribe@googlegroups.com
<mailto:raxml+unsubscribe@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

--
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/qXTeKk7ppGM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+unsubscribe@googlegroups.com.

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

--

Adrian Baez-Ortega
linkedin.com/in/baezortega

www.divulgatum.com

Alexey Kozlov

unread,

Sep 23, 2016, 7:36:00 AM9/23/16

to ra...@googlegroups.com

Hi Adrian,

> I've been so naive as to forget to mention the most important thing about my data: it's a coding sequence containing
> only somatic mutations from cancer samples (it's a very old contagious cancer lineage with an actual phylogeny), so
> probably the partitioning doesn't apply in this context, because negative selection is extremely weak or nonexistent in
> cancer, so we don't expect different mutation rates for each codon position. My mistake was trying to exactly replicate
> another pipeline that was conceived for flu data!

OK, but still that's something you can evaluate experimentally, i.e. if mutation rates are the same across all codon
positions, then "-M" should give you partition trees of similar length.

> What I don't know if I should still keep the HKY85 model, after all. I don't know if it depends more on selection or
> just on the fact that it is a coding sequence.

HKY85 is a special case of GTR, so the reason for using it is to avoid overfitting in cases when you don't have enough
data to estimate independent GTR rates. If still in doubt, you can use tools like jModelTest to perform formal model
selection.

Best,
Alexey

> >> On 21 September 2016 at 21:19, Alexey Kozlov <alexei...@gmail.com <mailto:alexei...@gmail.com> <javascript:>

> >> raxml+un...@googlegroups.com <mailto:raxml%2Bun...@googlegroups.com> <javascript:>

> >> <mailto:raxml%2Bu...@googlegroups.com <mailto:raxml%252Bu...@googlegroups.com> <javascript:>>
> >> <mailto:raxml+un...@googlegroups.com <mailto:raxml%2Bun...@googlegroups.com> <javascript:>

> >> <mailto:raxml%2Bu...@googlegroups.com <mailto:raxml%252Bu...@googlegroups.com> <javascript:>>>.

> >> to raxml+un...@googlegroups.com <mailto:raxml%2Bun...@googlegroups.com> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com <mailto:raxml%252Bu...@googlegroups.com> <javascript:>>.

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

> >> <https://groups.google.com/d/optout <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout <https://groups.google.com/d/optout>>>.
> >>
> >>
> >>
> >>
> >> --
> >> Adrian Baez-Ortega
> >> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega> <http://linkedin.com/in/baezortega

> <http://linkedin.com/in/baezortega>> <http://linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>>>
> >> www.divulgatum.com <http://www.divulgatum.com> <http://www.divulgatum.com> <http://www.divulgatum.com>

> >>
> >> --
> >> 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%2Bun...@googlegroups.com> <javascript:>
> >> <mailto:raxml+un...@googlegroups.com <mailto:raxml%2Bun...@googlegroups.com> <javascript:>>.

> >> For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout <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> <http://www.exelixis-lab.org>

>
> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com <mailto:raxml%2Bunsu...@googlegroups.com>>.

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

> 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/qXTeKk7ppGM/unsubscribe

> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
> To unsubscribe from this group and all its topics, send an email to raxml+un...@googlegroups.com

> <mailto:raxml%2Bunsu...@googlegroups.com>.

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

> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>

> www.divulgatum.com <http://www.divulgatum.com>

>
> --
> 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>.

Alexandros Stamatakis

unread,

Sep 23, 2016, 7:58:31 AM9/23/16

to ra...@googlegroups.com

sorry getting old, now I remember, but you might want to run modeltest
anyway,

alexis

On 23.09.2016 13:35, Adrián Báez Ortega wrote:
> Hi Alexis,
>
> Sorry, what I mean is not that the sequences only consist of mutated
> bases, but that they are coding sequences (concatenations of many CDS
> sequences) where the only mutations present are somatic mutations (so,
> no germline). We discussed this kind of data some weeks ago and you
> concluded that it's better not to use asc. correction bias if I have the
> full sequences.
>
> Many thanks,
> Adrian
>
>
> On 23 September 2016 at 12:29, Alexandros Stamatakis
> <alexandros...@gmail.com

> <mailto:alexei...@gmail.com

> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>>
> <javascript:>
>

> >> <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com> <mailto:alexei...@gmail.com

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com>
> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>> <javascript:>>
> >> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bun...@googlegroups.com>
> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com>
> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>> <javascript:>>>.

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>> <javascript:>
> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com>
> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>> <javascript:>>.

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

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>> <javascript:>
> >> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bun...@googlegroups.com>
> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>> <javascript:>>.

> >> For more options, visit
> https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>>
> <https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <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>
> <http://www.exelixis-lab.org>
> <http://www.exelixis-lab.org>
>
> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>>.

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

> 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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>
>
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe

> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>.
> To unsubscribe from this group and all its topics, send an
> email to
> raxml+un...@googlegroups.com

> <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>.

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

> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>

> www.divulgatum.com <http://www.divulgatum.com>
> <http://www.divulgatum.com>
>

> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>>.

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

> --
> 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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
> To unsubscribe from this group and all its topics, send an email to

> raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>.
>
>
>
>
> --
> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> www.divulgatum.com <http://www.divulgatum.com>
>
> --
> 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>.

Adrián Báez Ortega

unread,

Sep 23, 2016, 8:02:11 AM9/23/16

to ra...@googlegroups.com

Hi Alexis,

I'll definitely do that and see. When jModelTest gives me the best model for the data (I bet it'll say GTRGAMMA), should I use that same model for both the tree construction and the marginal reconstruction?

Thank you,

Adrian

On 23 September 2016 at 12:58, Alexandros Stamatakis <alexandros...@gmail.com> wrote:

sorry getting old, now I remember, but you might want to run modeltest anyway,

alexis

On 23.09.2016 13:35, Adrián Báez Ortega wrote:

Hi Alexis,

Sorry, what I mean is not that the sequences only consist of mutated
bases, but that they are coding sequences (concatenations of many CDS
sequences) where the only mutations present are somatic mutations (so,
no germline). We discussed this kind of data some weeks ago and you
concluded that it's better not to use asc. correction bias if I have the
full sequences.

Many thanks,
Adrian

On 23 September 2016 at 12:29, Alexandros Stamatakis

<alexandros.stamatakis@gmail.com

it, send an email to raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>
<mailto:raxml%2Bunsubscribe@googlegroups.com
<mailto:raxml%252Bunsubscribe@googlegroups.com>>
<mailto:raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>
<mailto:raxml%2Bunsubscribe@googlegroups.com
<mailto:raxml%252Bunsubscribe@googlegroups.com>>>.

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

--
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/qXTeKk7ppGM/unsubscribe
<https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>

<https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe

<https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>.
To unsubscribe from this group and all its topics, send an
email to

raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>
<mailto:raxml%2Bunsubscribe@googlegroups.com
<mailto:raxml%252Bunsubscribe@googlegroups.com>>.

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

--
Adrian Baez-Ortega
linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
<http://linkedin.com/in/baezortega
<http://linkedin.com/in/baezortega>>
www.divulgatum.com <http://www.divulgatum.com>
<http://www.divulgatum.com>

--
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+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>
<mailto:raxml+unsubscribe@googlegroups.com

<mailto:raxml%2Bunsubscribe@googlegroups.com>>.

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

--
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/qXTeKk7ppGM/unsubscribe
<https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
To unsubscribe from this group and all its topics, send an email to

raxml+unsubscribe@googlegroups.com
<mailto:raxml%2Bunsubscribe@googlegroups.com>.

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

--
Adrian Baez-Ortega
linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
www.divulgatum.com <http://www.divulgatum.com>

--
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+unsubscribe@googlegroups.com
<mailto:raxml+unsubscribe@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

--
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/qXTeKk7ppGM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+unsubscribe@googlegroups.com.

Alexandros Stamatakis

unread,

Sep 23, 2016, 8:07:15 AM9/23/16

to ra...@googlegroups.com

On 23.09.2016 14:02, Adrián Báez Ortega wrote:
> Hi Alexis,
>
> I'll definitely do that and see.

:-)

> When jModelTest gives me the best model
> for the data (I bet it'll say GTRGAMMA), should I use that same model
> for both the tree construction and the marginal reconstruction?

yes

>
> Thank you,
> Adrian
>
>
>
> On 23 September 2016 at 12:58, Alexandros Stamatakis
> <alexandros...@gmail.com

> <mailto:alexandros...@gmail.com>> wrote:
>
> sorry getting old, now I remember, but you might want to run
> modeltest anyway,
>
> alexis
>
> On 23.09.2016 13:35, Adrián Báez Ortega wrote:
>
> Hi Alexis,
>
> Sorry, what I mean is not that the sequences only consist of mutated
> bases, but that they are coding sequences (concatenations of
> many CDS
> sequences) where the only mutations present are somatic
> mutations (so,
> no germline). We discussed this kind of data some weeks ago and you
> concluded that it's better not to use asc. correction bias if I
> have the
> full sequences.
>
> Many thanks,
> Adrian
>
>
> On 23 September 2016 at 12:29, Alexandros Stamatakis

> <alexandros...@gmail.com
> <mailto:alexandros...@gmail.com>
> <mailto:alexandros...@gmail.com

> <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com>
>
> <mailto:alexei...@gmail.com

> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>
> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>>>
> <javascript:>
>
> >> <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com>
> <mailto:alexei...@gmail.com

> <mailto:alexei...@gmail.com>> <mailto:alexei...@gmail.com

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>
> <mailto:raxml%252Bun...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>> <javascript:>

> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com>
> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>>

> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>
> <mailto:raxml%25252Bu...@googlegroups.com
> <mailto:raxml%2525252Bu...@googlegroups.com>>> <javascript:>>

> >>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bun...@googlegroups.com>
> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>>

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>
> <mailto:raxml%252Bun...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>> <javascript:>

> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com>
> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>>

> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>
> <mailto:raxml%25252Bu...@googlegroups.com
> <mailto:raxml%2525252Bu...@googlegroups.com>>> <javascript:>>>.

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>
> <mailto:raxml%252Bun...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>> <javascript:>

> >> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252Bu...@googlegroups.com>
> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>>

> <mailto:raxml%252Bu...@googlegroups.com
> <mailto:raxml%25252Bu...@googlegroups.com>
> <mailto:raxml%25252Bu...@googlegroups.com
> <mailto:raxml%2525252Bu...@googlegroups.com>>> <javascript:>>.

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>
> <mailto:raxml%252Bun...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>> <javascript:>

> >> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bun...@googlegroups.com>
> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>>

> <mailto:raxml%2Bun...@googlegroups.com
> <mailto:raxml%252Bun...@googlegroups.com>
> <mailto:raxml%252Bun...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>> <javascript:>>.

> 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> <http://www.exelixis-lab.org>
> <http://www.exelixis-lab.org>
> <http://www.exelixis-lab.org>
>
> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>

> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>
> <mailto:raxml%252Buns...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>>

> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>

> <mailto:raxml%252Buns...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>>>.

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

> 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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>
>
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>
>
>
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>
>
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe

> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>>.
> To unsubscribe from this group and all its topics,
> send an
> email to
> raxml+un...@googlegroups.com

> <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>

> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>
> <mailto:raxml%252Buns...@googlegroups.com
> <mailto:raxml%25252Bun...@googlegroups.com>>>.

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

> Adrian Baez-Ortega
> linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>>

> www.divulgatum.com <http://www.divulgatum.com>
> <http://www.divulgatum.com>
> <http://www.divulgatum.com>
>

> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com
> <mailto:raxml%252Buns...@googlegroups.com>>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com

> <mailto:raxml%252Buns...@googlegroups.com>>>.

> For more options, visit
> https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <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>
> <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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>
>
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>.
> To unsubscribe from this group and all its topics, send an
> email to

> raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>
> <mailto:raxml%2Bunsu...@googlegroups.com

> <mailto:raxml%252Buns...@googlegroups.com>>.

> For more options, visit https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>>.
>
>
>
>
> --
> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>
> www.divulgatum.com <http://www.divulgatum.com>
> <http://www.divulgatum.com>
>
> --
> 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%2Bunsu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com

> <mailto:raxml%2Bunsu...@googlegroups.com>>.

> For more options, visit https://groups.google.com/d/optout
> <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>
>
> --
> 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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
> To unsubscribe from this group and all its topics, send an email to

> raxml+un...@googlegroups.com
> <mailto:raxml%2Bunsu...@googlegroups.com>.

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

> 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>.

Adrian Baez-Ortega

unread,

Sep 28, 2016, 9:48:24 AM9/28/16

to raxml

Hi Alexis,

I've had another problem while running the marginal reconstruction of ancestral states. After reading the alignment, RAxML prinst this:

You are trying to infer per site likelihoods or ancestral states or
do calculations with an ascertainment bias correction
on an alignment containing 5860 sites consisting only of undetermined
characters. Please remove them first and then re-run RAxML!

As I said before, I am using a concatenation of coding sequences, so I can't remove a single base if I want to infer the mutation consequences and the genes where they lie aftwerwards. I tried previously to run a much smaller example (viral data) adding lots of N's at the end of the sequences, and RAxML didn't complain about it.

So, I don't know if there's a way to trick RAxML into accepting those undetermined characters, or what's the best way to go around the issue otherwise. I was wondering... if I replaced all these N's with the same base (say, A) in all my samples, would that have any effect on the reconstruction? (Please forgive my naivety!) I obviously don't have any mutations in those bases, and I hope not to have any in the codons they belong to.

Thanks for all your help! :)

Best,
Adrian

> <mailto:raxml%2Bu...@googlegroups.com>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>
> <mailto:raxml%252B...@googlegroups.com
> <mailto:raxml%25252Bunsubscribe@googlegroups.com>>>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bu...@googlegroups.com>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>
> <mailto:raxml%252B...@googlegroups.com
> <mailto:raxml%25252Bunsubscribe@googlegroups.com>>>>.

> <mailto:raxml%2Bu...@googlegroups.com>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>
> <mailto:raxml%252B...@googlegroups.com
> <mailto:raxml%25252Bunsubscribe@googlegroups.com>>>.

> <mailto:raxml%2Bu...@googlegroups.com>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bu...@googlegroups.com>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>>>.

> <mailto:raxml%2Bu...@googlegroups.com>
> <mailto:raxml%2Bu...@googlegroups.com
> <mailto:raxml%252B...@googlegroups.com>>.

> For more options, visit https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>
> <https://groups.google.com/d/optout
> <https://groups.google.com/d/optout>>.
>
>
>
>
> --
> Adrian Baez-Ortega
> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>
> www.divulgatum.com <http://www.divulgatum.com>
> <http://www.divulgatum.com>
>
> --
> 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%2Bu...@googlegroups.com>
> <mailto:raxml+un...@googlegroups.com
> <mailto:raxml%2Bu...@googlegroups.com>>.

> For more options, visit https://groups.google.com/d/optout
> <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>
>
> --
> 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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>.
> To unsubscribe from this group and all its topics, send an email to
> raxml+un...@googlegroups.com

> <mailto:raxml%2Bu...@googlegroups.com>.

Alexey Kozlov

unread,

Sep 28, 2016, 5:58:01 PM9/28/16

to ra...@googlegroups.com

Hi Adrian,

RAxML complains about "empty" columns, i.e. columns which contain *only* gaps/Ns. Those are not allowed (and
automatically removed before "regular" tree inference) for a reason - although such columns do not have any signal, they
would affect branch length estimates. Same thing will happen if you would change them to As (=introducing extra constant
sites), so I don't think it's a good solution.

On the other hand, I can see your problem with preserving original base positions.

So what you can do is to write a simple script which will pre-process the alignment by removing all gap-only columns and
will write out the "original" index for each column in the reduced alignment, e.g.:

1 2 4 5 7 8 9 ...

if columns 3 and 6 were removed.

After performing the ancestral state reconstruction, you can then use this mapping to put ancestral state probabilities
in their original positions.

Hope this helps,
Alexey

> I've had another problem while running the marginal reconstruction of ancestral states. After reading the alignment,
> RAxML prinst this:
>
> You are trying to infer per site likelihoods or ancestral states or
> do calculations with an ascertainment bias correction
> on an alignment containing 5860 sites consisting only of undetermined
> characters. Please remove them first and then re-run RAxML!
>
> As I said before, I am using a concatenation of coding sequences, so I can't remove a single base if I want to infer the
> mutation consequences and the genes where they lie aftwerwards. I tried previously to run a much smaller example (viral
> data) adding lots of N's at the end of the sequences, and RAxML didn't complain about it.

>
> So, I don't know if there's a way to trick RAxML into accepting those undetermined characters, or what's the best way to
> go around the issue otherwise. I was wondering... if I replaced all these N's with the same base (say, A) in all my
> samples, would that have any effect on the reconstruction? (Please forgive my naivety!) I obviously don't have any
> mutations in those bases, and I hope not to have any in the codons they belong to.

> On Friday, 23 September 2016 13:07:15 UTC+1, Alexis wrote:
>
>
>
> On 23.09.2016 14:02, Adrián Báez Ortega wrote:
> > Hi Alexis,
> >
> > I'll definitely do that and see.
>
> :-)
>
> > When jModelTest gives me the best model
> > for the data (I bet it'll say GTRGAMMA), should I use that same model
> > for both the tree construction and the marginal reconstruction?
>
> yes
>
> >
> > Thank you,
> > Adrian
> >
> >
> >
> > On 23 September 2016 at 12:58, Alexandros Stamatakis

> > <alexandros...@gmail.com <javascript:>

> > <mailto:alexandros...@gmail.com <javascript:>>> wrote:
> >
> > sorry getting old, now I remember, but you might want to run
> > modeltest anyway,
> >
> > alexis
> >
> > On 23.09.2016 13:35, Adrián Báez Ortega wrote:
> >
> > Hi Alexis,
> >
> > Sorry, what I mean is not that the sequences only consist of mutated
> > bases, but that they are coding sequences (concatenations of
> > many CDS
> > sequences) where the only mutations present are somatic
> > mutations (so,
> > no germline). We discussed this kind of data some weeks ago and you
> > concluded that it's better not to use asc. correction bias if I
> > have the
> > full sequences.
> >
> > Many thanks,
> > Adrian
> >
> >
> > On 23 September 2016 at 12:29, Alexandros Stamatakis

> > <alexandros...@gmail.com <javascript:>
> > <mailto:alexandros...@gmail.com <javascript:>>
> > <mailto:alexandros...@gmail.com <javascript:>

> > <alexei...@gmail.com <javascript:>
> > <mailto:alexei...@gmail.com <javascript:>> <mailto:alexei...@gmail.com <javascript:>
> > <mailto:alexei...@gmail.com <javascript:>>>
> > <mailto:alexei...@gmail.com <javascript:>
> > <mailto:alexei...@gmail.com <javascript:>>
> >
> > <mailto:alexei...@gmail.com <javascript:>

> > raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>

> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>
> > <mailto:raxml%25252Bun...@googlegroups.com <javascript:>>>>
> > <mailto:raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>

> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>>>.
> > To unsubscribe from this group and all its topics,
> > send an
> > email to

> > raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>

> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>

> > www.divulgatum.com <http://www.divulgatum.com> <http://www.divulgatum.com>
> > <http://www.divulgatum.com>
> > <http://www.divulgatum.com>
> >

> > --
> > 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 <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>

> > <mailto:raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>

> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>>.

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

> > <https://groups.google.com/d/optout <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> <http://www.exelixis-lab.org>
> > <http://www.exelixis-lab.org>
> >

> > raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>

> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>.

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

> > <https://groups.google.com/d/optout <https://groups.google.com/d/optout>>>.
> >
> >
> >
> >
> > --
> > Adrian Baez-Ortega
> > linkedin.com/in/baezortega <http://linkedin.com/in/baezortega> <http://linkedin.com/in/baezortega
> <http://linkedin.com/in/baezortega>>
> > <http://linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> > <http://linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>>>
> > www.divulgatum.com <http://www.divulgatum.com> <http://www.divulgatum.com>
> > <http://www.divulgatum.com>
> >
> > --
> > 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 <javascript:>

> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>
> > <mailto:raxml+un...@googlegroups.com <javascript:>

> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>>.

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

> > <https://groups.google.com/d/optout <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> <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/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>
> > <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe
> <https://groups.google.com/d/topic/raxml/qXTeKk7ppGM/unsubscribe>>.
> > To unsubscribe from this group and all its topics, send an email to

> > raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>.

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

> > <https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
> >
> >
> >
> > --
> > 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 <javascript:>
> > <mailto:raxml+un...@googlegroups.com <javascript:>>.

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

Adrian Baez-Ortega

unread,

Sep 29, 2016, 4:36:22 AM9/29/16

to raxml

Hi Alexey,

Many thanks for your reply. I see, I could do something like that, with the downside that it would need to go through every sequence to check if each position changes or not. On the other hand, these 5860 undetermined sites make up ~0.018% of my original sequence (I have ~28K mutations imposed onto a 32Gb concatenated sequence), so I don't think that turning them into A's would make a big difference, am I right?

Best,
Adrian

> > <mailto:raxml%25252Bunsubscribe@googlegroups.com <javascript:>>>>

> > <mailto:raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>>
> > <mailto:raxml%2Bu...@googlegroups.com <javascript:>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>>
> > <mailto:raxml%252B...@googlegroups.com <javascript:>

> > <mailto:raxml%25252Bunsubscribe@googlegroups.com <javascript:>>>>>.

> > <mailto:raxml%25252Bunsubscribe@googlegroups.com <javascript:>>>>.

Adrian Baez-Ortega

unread,

Sep 29, 2016, 6:36:38 AM9/29/16

to raxml

Sorry, I made a mistake. Of course I don't have 32 Gb of coding sequence, but 32 Mb!

Adrian

...

Alexandros Stamatakis

unread,

Sep 29, 2016, 6:52:15 AM9/29/16

to ra...@googlegroups.com

If you start a normal search with raxml it will complain about the
undetermined sites and also write to file a reduced partition and
alignment file from which the undetermined sites have been removed (just
look at the terminal output or the RAxML_info file) you can then use
this reduced alignment for ancest. state reconstr.

alexis

> <http://linkedin.com/in/baezortega>>
> > > <
>
> ...

>
> --
> 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.

Alexey Kozlov

unread,

Sep 29, 2016, 8:29:50 AM9/29/16

to ra...@googlegroups.com

Hi Adrian,

> Sorry, I made a mistake. Of course I don't have 32 Gb of coding sequence, but 32 Mb!

thanks for clarifying, I was already wondering what kind of beast you're working with :)

> Many thanks for your reply. I see, I could do something like that, with the downside that it would need to go
> through every sequence to check if each position changes or not. On the other hand, these 5860 undetermined sites
> make up ~0.018% of my original sequence (I have ~28K mutations imposed onto a 32Gb concatenated sequence), so I
> don't think that turning them into A's would make a big difference, am I right?

OK, then maybe you're right and there will be no major difference on your dataset. You can easily test it by running
"raxml -f e" on the modified alignment (with all-Ns -> all-As) and check if there will be a noticeable difference in
branch length compared to the original one.

Otherwise, you can follow Alexis' suggestion above and use empty column indexes reported by RAxML, so that you don't
need to go through alignment yourself. You would still need to back-map the inferred ancestral probs to the original
sequence, however.

Best,
Alexey

> <http://linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>>
> > > <
>
> ...

>
> --
> 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.

Adrian Baez-Ortega

unread,

Sep 29, 2016, 12:19:39 PM9/29/16

to raxml

Hi Alexey,

Actually, I only need the sequences to stay unaltered (in terms of number of sites) for the marginal reconstruction, but for the tree construction, the original sequences are processed by RAxML and, if I understand correctly, the analysis is done on the version with undetermined sites removed. (I'm not trying to adjusting branch lengths a posteriori for now.) So I think that's ideal in this case – the tree construction routine receives the original sequences with Ns and reduces them, while the marginal reconstruction routine receives the sequences with Ns replaced by As.

Thank you!
Adrian

Alexandros Stamatakis

unread,

Sep 29, 2016, 5:51:39 PM9/29/16

to ra...@googlegroups.com

Hi Adrian,

> Actually, I only need the sequences to stay unaltered (in terms of
> number of sites) for the marginal reconstruction, but for the tree
> construction, the original sequences are processed by RAxML and, if I
> understand correctly, the analysis is done on the version with
> undetermined sites removed.

No, RAxML generates a file with undetermined sites removed, which can be
used in another run.

> (I'm not trying to adjusting branch lengths
> a posteriori for now.) So I think that's ideal in this case – the tree
> construction routine receives the original sequences with Ns and reduces
> them, while the marginal reconstruction routine receives the sequences
> with Ns replaced by As.

But this will alter your results for the ancestral reconstruction at
those positions.

Alexis

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

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

> <https://groups.google.com/d/optout>.
>
> --

> 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.

Adrián Báez Ortega

unread,

Sep 30, 2016, 6:11:16 AM9/30/16

to ra...@googlegroups.com

Hi Alexis,

Thanks for clarifying! So, for the tree search, do you think that it is better (I guess in terms of branch length estimates) re-running RAxML with the reduced file (without undetermined sites) that it generates, instead of just taking the output from the initial run?

As for the ancestral reconstruction, do you think that if I replaced a column that was originally 'N' in every sample with an 'A' in every sample, the ancestral reconstruction could give me ancestral states that are not 'A'? Is that the problem that you are referring to?

Many thanks,

Adrian

Alexandros Stamatakis

unread,

Sep 30, 2016, 8:35:39 AM9/30/16

to ra...@googlegroups.com

On 30.09.2016 13:11, Adrián Báez Ortega wrote:
> Hi Alexis,
>

> Thanks for clarifying! So, for the tree search, do you think that it is
> better (I guess in terms of branch length estimates) re-running RAxML
> with the reduced file (without undetermined sites) that it generates,
> instead of just taking the output from the initial run?

yes.

> As for the ancestral reconstruction, do you think that if I replaced a
> column that was originally 'N' in every sample with an 'A' in every
> sample, the ancestral reconstruction could give me ancestral states that
> are not 'A'? Is that the problem that you are referring to?

It is very unlikely

Alexis

>
> Many thanks,
> Adrian
>
> On 29 September 2016 at 22:51, Alexandros Stamatakis
> <alexandros...@gmail.com

> <mailto:alexandros...@gmail.com> <javascript:>
> > > > <mailto:alexandros...@gmail.com

> <mailto:alexandros...@gmail.com> <javascript:>
> > > > <mailto:alexandros...@gmail.com
> <mailto:alexandros...@gmail.com>
> <javascript:>>
> > > > <mailto:alexandros...@gmail.com
> <mailto:alexandros...@gmail.com> <javascript:>
> > > > <mailto:alexandros...@gmail.com

> <mailto:alexei...@gmail.com> <javascript:>
> > > > <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com> <javascript:>>
> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>
> <javascript:>
> > > > <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com> <javascript:>>>
> > > >
> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>
> <javascript:>
> > > > <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com> <javascript:>>
> > > >
> > > >
> <mailto:alexei...@gmail.com <mailto:alexei...@gmail.com>
> <javascript:>
> > > > <mailto:alexei...@gmail.com
> <mailto:alexei...@gmail.com>

> linkedin.com/in/baezortega <http://linkedin.com/in/baezortega>
> www.divulgatum.com <http://www.divulgatum.com>
>

Adrian Baez-Ortega

unread,

Sep 30, 2016, 10:35:13 AM9/30/16

to raxml

Hi Alexis,

Thank you for your answers. I just have another quick doubt, since I haven't been able to find the answer in this group.

I was looking in the manual for a "-f ..." option that would allow me to reduce the input file, removing the undetermined sites without actually running the tree search (I don't know if "-f c" would do, or maybe a very fast search with "-f E"?), and I came across this interesting one:

-f T

do final thorough optimization of ML tree from rapid bootstrap search in standalone mode

This option allows to do a more thorough tree search that uses less lazy, i.e., more exhaustive SPR moves, in stand-alone mode.

This algorithm is typically executed in the very end of a search done via -f a.

What I'd like to clarify is whether the last sentence means that "-f T" should be applied to the output of "-f a", or if it is actually already included in the "-f a" algorithm. I know that "-f a" includes rapid bootstrap, fast ML search, slow ML search, thorough ML search and final ML optimization, so although bootstrap is performed at the beginning I would dare to say that it is used for the "thorough ML search" in "-f a" and therefore it is not necessary to explicitly run it?

I hope I will eventually stop asking stuff. :-)

Many thanks,
Adrian

Alexandros Stamatakis

unread,

Sep 30, 2016, 10:39:00 AM9/30/16

to ra...@googlegroups.com

just run the tree search and then kill the process once the file has
been written,

alexis

On 30.09.2016 17:35, Adrian Baez-Ortega wrote:
> Hi Alexis,
>
> Thank you for your answers. I just have another quick doubt, since I
> haven't been able to find the answer in this group.
>
> I was looking in the manual for a "-f ..." option that would allow me to
> reduce the input file, removing the undetermined sites without actually
> running the tree search (I don't know if "-f c" would do, or maybe a
> very fast search with "-f E"?), and I came across this interesting one:
>
> -f T
>
> do final thorough optimization of ML tree from rapid bootstrap search in
> standalone mode
>
> This option allows to do a more thorough tree search that uses less
> lazy, i.e., more exhaustive SPR moves, in stand-alone mode.
>

> *This algorithm is typically executed in the very end of a search done
> via -f a. *

> > <alexandros...@gmail.com <javascript:>

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

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

> <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>

Reply all

Reply to author

Forward