Question about the exact steps of the analysis using "-m PROTGAMMAAUTO", "-x" and " -f a"

[Quanwei Zhang]

Hello:

I am using the following command to generate the phylogenetic tree. Since I used "-m PROTGAMMAAUTO",  "-x" and " -f a". I wonder what RAxML exactly does in each step.

raxmlHPC-PTHREADS-SSE3 -T 15 -n test1 -s Alignment.fa -o Human,Chimpanzee -p 12345 -m PROTGAMMAAUTO -f a -N 100 -x 12345

Here is my understanding. Is it correct?
Firstly, with "-p 12345" RAxML generated a parsimony tree based on the alignments.
Secondly, with "-m PROTGAMMAAUTO" RAxML compares the empirical base frequencies(in our alignments) with those coming with the protein substitution model (e..g., JTT, WAG). And then chose the best protein substitution model.
Thirdly, with "-x" and " -f a"  RAxML will generate 100 ("-N 100") trees (fast boostrapping) and for each tree perform maximum likelihood.
Finally, RAxML select the one with the maximum likelihood

Thank you. BTW, is my command is accurate and suggested to generate a more reliable phylogenetic tree?

Best
Quanwei

[Alexandros Stamatakis]

Dear Quanwei,

> I am using the following command to generate the phylogenetic tree.
> Since I used "-m PROTGAMMAAUTO",  "-x" and " -f a". I wonder what RAxML
> exactly does in each step.
>
> raxmlHPC-PTHREADS-SSE3 -T 15 -n test1 -s Alignment.fa -o
> Human,Chimpanzee -p 12345 -m PROTGAMMAAUTO -f a -N 100 -x 12345
>
> Here is my understanding. Is it correct?
> Firstly, with "-p 12345" RAxML generated a parsimony tree based on the
> alignments.

Not only one parsimony tree, -p is the random number seed for
generatimng several randomized stepwise addition parsimony trees.

> Secondly, with "-m PROTGAMMAAUTO" RAxML compares the empirical base
> frequencies(in our alignments) with those coming with the protein
> substitution model (e..g., JTT, WAG). And then chose the best protein
> substitution model.

Not exactly, it compares the likelihoods of all prot subst. models and
selects the best fitting one.

> Thirdly, with "-x" and " -f a"  RAxML will generate 100 ("-N 100") trees
> (fast boostrapping) and for each tree perform maximum likelihood.
> Finally, RAxML select the one with the maximum likelihood

Correct.

> Thank you. BTW, is my command is accurate and suggested to generate a
> more reliable phylogenetic tree?

You can always do a thorough bootstrap search, check section XI of the
manual starting at page 53:

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

Alexis

>
> Best
> Quanwei
>
> --
> 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

[Quanwei Zhang]

Hi Alexis:

Thank you for your explanation. I am still not clear about some points (shown in red below). Would you please further explain?

I am using the following command to generate the phylogenetic tree. Since I used "-m PROTGAMMAAUTO",  "-x" and " -f a". I wonder what RAxML exactly does in each step.

raxmlHPC-PTHREADS-SSE3 -T 15 -n test1 -s Alignment.fa -o Human,Chimpanzee -p 12345 -m PROTGAMMAAUTO -f a -N 100 -x 12345

Here is my understanding. Is it correct?
Firstly, with "-p 12345" RAxML generated a parsimony tree based on the alignments.

Not only one parsimony tree, -p is the random number seed for generatimng several randomized stepwise addition parsimony trees.

In my understanding, if I do not use boostrapping, "-N" indicates the number of pasimony starting tree, that RAxML will generate. But in my command, "-N" tells RAxML the number for boostrapping, then how many parsimony trees will  RAxML generate by my command?

Secondly, with "-m PROTGAMMAAUTO" RAxML compares the empirical base frequencies(in our alignments) with those coming with the protein substitution model (e..g., JTT, WAG). And then chose the best protein substitution model.

Not exactly, it compares the likelihoods of all prot subst. models and selects the best fitting one.

Thirdly, with "-x" and " -f a"  RAxML will generate 100 ("-N 100") trees (fast boostrapping) and for each tree perform maximum likelihood.
Finally, RAxML select the one with the maximum likelihood

Correct.

Thank you. BTW, is my command is accurate and suggested to generate a more reliable phylogenetic tree?

You can always do a thorough bootstrap search, check section XI of the manual starting at page 53:

For "a thorough bootstrap search", isn't that what "-f a" indicates? I can increase the number of "-N" (or use "autoMRE") for a more thorough bootstrap search, right?

Many thanks

Best

Quanwei

2017-10-12 3:03 GMT-04:00 Alexandros Stamatakis <alexandros...@gmail.com>:

Dear Quanwei,

I am using the following command to generate the phylogenetic tree. Since I used "-m PROTGAMMAAUTO",  "-x" and " -f a". I wonder what RAxML exactly does in each step.

raxmlHPC-PTHREADS-SSE3 -T 15 -n test1 -s Alignment.fa -o Human,Chimpanzee -p 12345 -m PROTGAMMAAUTO -f a -N 100 -x 12345

Here is my understanding. Is it correct?
Firstly, with "-p 12345" RAxML generated a parsimony tree based on the alignments.

Not only one parsimony tree, -p is the random number seed for generatimng several randomized stepwise addition parsimony trees.

Secondly, with "-m PROTGAMMAAUTO" RAxML compares the empirical base frequencies(in our alignments) with those coming with the protein substitution model (e..g., JTT, WAG). And then chose the best protein substitution model.

Not exactly, it compares the likelihoods of all prot subst. models and selects the best fitting one.

Thirdly, with "-x" and " -f a"  RAxML will generate 100 ("-N 100") trees (fast boostrapping) and for each tree perform maximum likelihood.
Finally, RAxML select the one with the maximum likelihood

Correct.

Thank you. BTW, is my command is accurate and suggested to generate a more reliable phylogenetic tree?

You can always do a thorough bootstrap search, check section XI of the manual starting at page 53:

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

Alexis

Best
Quanwei

--
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

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/8bF0Uzjfssw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+unsubscribe@googlegroups.com.

[Alexandros Stamatakis]

On 12.10.2017 20:35, Quanwei Zhang wrote:
> Hi Alexis:
>
> Thank you for your explanation. I am still not clear about some points
> (shown in red below). Would you please further explain?
>
> I am using the following command to generate the phylogenetic tree.
> Since I used "-m PROTGAMMAAUTO",  "-x" and " -f a". I wonder what
> RAxML exactly does in each step.
>
> raxmlHPC-PTHREADS-SSE3 -T 15 -n test1 -s Alignment.fa -o
> Human,Chimpanzee -p 12345 -m PROTGAMMAAUTO -f a -N 100 -x 12345
>
> Here is my understanding. Is it correct?
> Firstly, with "-p 12345" RAxML generated a parsimony tree based on
> the alignments.
>
>
> Not only one parsimony tree, -p is the random number seed for
> generatimng several randomized stepwise addition parsimony trees.
>
> In my understanding, if I do not use boostrapping, "-N" indicates the
> number of pasimony starting tree, that RAxML will generate. But in my
> command, "-N" tells RAxML the number for boostrapping, then how many
> parsimony trees will RAxML generate by my command?

It's a bit complicated for the rapid BS, it will generate 10 starting
trees for the BS (1/10 of N) and 20 for the ML search (1/5 of N)

>
> Secondly, with "-m PROTGAMMAAUTO" RAxML compares the empirical base
> frequencies(in our alignments) with those coming with the protein
> substitution model (e..g., JTT, WAG). And then chose the best
> protein substitution model.
>
>
> Not exactly, it compares the likelihoods of all prot subst. models and
> selects the best fitting one.
>
> Thirdly, with "-x" and " -f a"  RAxML will generate 100 ("-N 100")
> trees (fast boostrapping) and for each tree perform maximum likelihood.
> Finally, RAxML select the one with the maximum likelihood
>
>
> Correct.
>
> Thank you. BTW, is my command is accurate and suggested to generate
> a more reliable phylogenetic tree?
>
>
> You can always do a thorough bootstrap search, check section XI of the
> manual starting at page 53:
>
> For "a thorough bootstrap search", isn't that what "-f a" indicates? I
> can increase the number of "-N" (or use "autoMRE") for a more thorough
> bootstrap search, right?

No, please read the manual pages I have indicated.

Alexis

>
> Many thanks
>
> Best
> Quanwei
>
>
> 2017-10-12 3:03 GMT-04:00 Alexandros Stamatakis

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

> 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
>

> 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/8bF0Uzjfssw/unsubscribe

> <https://groups.google.com/d/topic/raxml/8bF0Uzjfssw/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>.