Silly question

[Patrick Sisler]

Hi, I'm running standard-RAxML - 8.2.12, and I'm pretty new at this but i'm getting this error: ```-bash: raxmlHPC-PTHREADS-SSE3: command not found``` when running this line: ```raxmlHPC-PTHREADS-SSE3 -f a -# 500 -m GTRGAMMA -p 12345 -x 12345 -s Helianthus_chloroplast.fasta -n Helianthus_chloro.tree```. Again I'm pretty new at this if someone could point me in the direction. 

[Alexandros Stamatakis]

Is RAxML located in your /bin directory?

If you are executing raxml in the directory where it is located try:

./raxmlHPC-PTHREADS-SSE3

Also, please switch to the latest version RAxML-NG:

https://github.com/amkozlov/raxml-ng

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>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/raxml/6aebc38c-30e9-4bf0-a3c7-6516e89edb87%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/6aebc38c-30e9-4bf0-a3c7-6516e89edb87%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology

www.exelixis-lab.org

[Patrick Sisler]

Thanks for the help I got it working, but I'm trying to use the -J option to get a custom consensus threshold. I'm using the argument -J MR T_50, but I keep getting this: The file  you want to open for reading does not exist, exiting ... When this argument is not present RAxML runs just fine. Any advise?

[Patrick Sisler]

Could you specify the -z, specifically the input file.

On Friday, June 21, 2019 at 12:54:45 PM UTC-4, Patrick Sisler wrote:

[Alexandros Stamatakis]

sorry I don't get the question,

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>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/b420236c-a7c0-448f-b1d6-7fb2cd8f05ad%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/b420236c-a7c0-448f-b1d6-7fb2cd8f05ad%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Grimm]

Am Montag, 8. Juli 2019 06:47:05 UTC+2 schrieb Alexis:

sorry I don't get the question,
 

I think I may, Alexi

Patrick, it seems you're trying to mix two different functions of RAxML.

With -f a you run a full analysis, this gives you, what we sometimes call, a "best-known" tree (outfile is: RAxML_besttree...) and performs a bootstrap analysis, in your case with 500 pseudoreplicates (RAxML_bootstrap...). -f a will automatically plot the BS support on the best-known tree (RAxML_branchsupport; ...branchlabelled).

The option "-n Helianthus_chloro.tree" just labels your file, but since you asked for -z and -J options, I guess, you want to read in a tree and map the support using a threshold?

-J just calculates a consensus, as a post-analysis wrapup (there are still many journals/editors/reviewers who like arbitrary cut-off rather than having the actual BS support on each branch; I don't see the point in that since, eg., a BS = 49 is not signifanctly worse than BS = 51).

And yes, you need to read in a tree sample for this option with -z; for instance the BS pseudoreplicate sample you generated with RAxML.

E.g.

raxmlHPC-PTHREADS-SSE3 -f a -# 500 -m GTRGAMMA -p 12345 -x 12345 -s Helianthus_chloroplast.fasta -n Helianthus

raxmlHPC - J T_xx -z RAxML_bootstrap.Helianthus -m GTRGAMMA -n Plot    

with xx being the threshold you want and you get a consensus tree showing only the branches that occured >= xx % of the BS pseudoreplicate trees.

Cheers, Guido

PS If you have branches with BS << 100, any consensus tree represents a poor means to sum this up, instead you should use consensus networks showing you the alternative competing topological scenarios; in case of chloroplast data the most common reason for BS << 100 is that your data combine plastid markers of different phylogenetic information content (divergence), it's not uncommon that coding genes and noncoding spacer tell different stories. To get a consensus network you can eg. just open the RAxML_bootstrap tree sample with SplitsTree (www.splitstree.org)

[I posted a walkthrough on my Res.I.P. blog, still using the classic RAxML you used, but pretty much the same with the new RAxML-ng: How did I do it — a short guide to a nice graph)

[Patrick Sisler]

Thanks for your advise Guido. The -z imput tree is suppose to be an unrooted correct? Is there a way to check if a tree is unrooted? Because I ran RAxML first without specifying any outgroups or a some of the other output files unrooted? For this RAxML run, I've specified my input file, my outgroups, my consensus threshold (50), my rapid bootstrap analysis with 500 pseudoreplicate, my model (GTRGAMA), my output file, but everytime I run with - J T_50 I get an error, 'Error, starting tree(s) will be ignored by rapid Bootstrapping'. Is it not possible to run all the processes at the same time? Also is the default consensus threshold 48%?

Again thanks for your help, it has been very useful

Patrick

[Alexandros Stamatakis]

On 09.07.19 03:16, Patrick Sisler wrote:
> Thanks for your advise Guido. The -z imput tree is suppose to be an
> unrooted correct? Is there a way to check if a tree is unrooted? Because
> I ran RAxML first without specifying any outgroups or a some of the
> other output files unrooted? For this RAxML run, I've specified my input
> file, my outgroups, my consensus threshold (50), my rapid bootstrap
> analysis with 500 pseudoreplicate, my model (GTRGAMA), my output file,
> but everytime I run with - J T_50 I get an error, 'Error, starting
> tree(s) will be ignored by rapid Bootstrapping'. Is it not possible to
> run all the processes at the same time?

It is not.

> Also is the default consensus
> threshold 48%?

No, where did you get this number from?

Alexis

> <http://www.splitstree.org>)

> [I posted a walkthrough on my Res.I.P. blog, still using the classic

> RAxML you used, but pretty much the same with the new RAxML-ng: /How

> did I do it — a short guide to a nice graph

> <https://researchinpeace.blogspot.com/2018/02/how-did-i-do-it-short-guide-to-nice.html>)/

>
>
> --
> 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>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/81e49c1a-4a84-4e7c-81bc-92865f71519e%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/81e49c1a-4a84-4e7c-81bc-92865f71519e%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Grimm]

Am Dienstag, 9. Juli 2019 06:33:22 UTC+2 schrieb Alexis:

On 09.07.19 03:16, Patrick Sisler wrote:
> Thanks for your advise Guido. The -z imput tree is suppose to be an
> unrooted correct?

Yes. For the consensus tree rooting is irrelevant (like for anything else except ancestral state reconstruction). Rooting with an outgroup is a post-analysis modification of the inferred tree, a purely graphical one.

Like making a consensus tree...

 

Is it not possible to
> run all the processes at the same time?

It is not.

 ... since calculating a consensus tree is a post-analysis step. You need to make an according shell/bash (depending whether you run from the LINUX source code or the Windows-executable)

 

> Also is the default consensus
> threshold 48%?

No, where did you get this number from?

I suppose from my example. As the name implies, the "majority" rule tree includes all branches that are found in the majority of BS replicates. Since the dawn of phylogenetics, the default has hence been 50 (absolute majority) although e.g. in PAUP* one had a "LE50" option, which would relax the threshold, and in MrBayes you could use "allcompat" for the sumt command.

In plant phylogenetics, many like to feel safe with using 70, and regard all that is below (i.e. 69.999999999999 and smaller) as irrelevant lacking support.

But again, when you have branches with BS smaller than 100, it means the signal in your data is not unambiguous, and then the best option is to use a consensus network. For instance, you have A, B, and C all part of the same subtree (clade in a rooted tree)

if 50% of your distinct alignmentpattern favour A as sister to B, but the rest favour A sister to C, this may be represented in the BS replicate sample and you get a split support of e.g. 55 to 45 for A+B vs. A+C.

When you infer the consensus tree with a threshold of 60, it will show an A-B-C polytomy, which is not wrong but also doesn't tell the whole truth because your data didn't support B sister C.

With a threshold of 50, your consensus tree will have (A + B) + C, which is pretty wrong, since (A + C) + B is not that worse as alternative.

Reducing the threshold will change little, because 55 > 45, and any tree is strictly dichotomous.

A consensus network with a threshold of e.g. 33 will show both alternatives, hence, best captures your data/bootstrapping reality.

With a threshold of 50, your tree

 

> an email to ra...@googlegroups.com
> <mailto:ra...@googlegroups.com>.

[Patrick Sisler]

Thanks for the help, I think I have it all figured out now