Re: BSSVS using BEAST2

[Andrew Rambaut]

Dear Veera,

XML generated by BEAUti 1.7.x (or written using the tutorials) can only be run in BEAST 1.7.x. However, you shouldn't need to edit the XML - BEAUti will generate BSSVS models directly.

Best,
Andrew

On 14 Feb 2013, at 17:36, veera jaime <vee...@yahoo.com> wrote:

> HI Beast Users,
>
> I've performed a discrete phylogeographic analysis by using the BEAST2 package. I've generated the xml file using Beauti in BEAST2 and I've managed to view the results in SPREAD. But, now i would like to perform Bayes factor analysis using the BSSVS stochastic procedure to determine the most frequently invoke rates and to view it in SPREAD. Can anyone help me how to edit the xml input file generated from the Beauti from BEAST2 package for the BSSVS procedure? Can Beauti in BEAST2 package prepare the input file for BSSVS procedue? I've tried to edit it based on the tutorial provided for xml edit for BSSVS on the website for BEAST package version 1.7 but the xml formats seemed different between the versions and I've failed to run it on BEAST2. Please give me suggestions on how to get this done.
>
> Btw, I'm working on insects phylogeography and try to determine the diffusion patterns across geographical regions.
>
> Thanking in advance.
>
> Cheers,
>
> Veera
>
> --
> You received this message because you are subscribed to the Google Groups "beast-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
> To post to this group, send email to beast...@googlegroups.com.
> Visit this group at http://groups.google.com/group/beast-users?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

___________________________________________________________________
Andrew Rambaut
Institute of Evolutionary Biology University of Edinburgh
Ashworth Laboratories Edinburgh EH9 3JT
EMAIL - a.ra...@ed.ac.uk TEL - +44 131 6508624


--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

[veera jaime]

Dear Andrew,

Thank so much you for the prompt reply. I really appreciate it. If I understand this correctly from the SPREAD tutorial, I would require a BEAST log file with rate indicators to identify the well supported rates using Bayes factor test. When I use the trace log file generated from BEAST2 using xml generated from BEAUti to run SPREAD, the program fail to perform the Bayes factor analysis. May I know how can I get the log file with rate indicators using BEAST2? Is there any settings that i need to do in BEAUti prior running the xml  in BEAST2?

Cheers,

Veera

[Alexei Drummond]

Dear Veera,

BEAST2 was a complete re-write of BEAST1 to make some things better including:

(1) BEAUTI2 and BEAST2 now use the same XML input files so that models can easily be modified by BEAUTI at a later time.

(2) BEAST2 periodically logs the full state to disk which allows for checkpointing and easily continuing an analysis if it hasn't been run long enough.

(3) BEAST2 has an Add-on structure so that updates to add-ons can be made without an entire re-release of the whole software package.

(4) BEAUTI2 user interface is now programmed by XML templates rather than Java code allowing new models to be more easily added to the user interface so that XML editing is not needed as much.

However because BEAST2 was a re-write we are still working though the compatibilities with some of the auxiliary software that has developed around BEAST 1.x. Thanks for bringing this particular issue to our attention. If you could email your log file to me we should be able to fix this problem very quickly.

Cheers

Alexei

[Remco Bouckaert]

Hi Veera,

A quick fix to do BFs in SPREAD is to edit the log file as follows:

1. remove all comments -- those are lines starting with a hash -- from
the start of the file

2. rename the rateIndicator.location columns to indicator. This is
assuming you called your partition 'location', if it is something else,
there should be columns named rateIndicator.<partition name>. The
columns should be numbered now and called indicator1, indicator2,
indicator3, etc.

With these two changes, SPREAD should work as advertised. Let me know if
this doesn't.

Remco

[veera jaime]

Dear Alexei,

Thank you so much for giving concern over my predicaments here. Indeed, BEAST2 is a great package with more user friendly features in various ways especially when it comes to xml file preparation via GUI in BEAUti. Please find attached the log files from two separate runs namely mdataset5disBEAST and mdataset5disBEAST2, I've also attached the log combined file of these two log files (mdataset5disBEASTcombo) without any burn-in settings. I'm so glad that this could be fixed, thank you so much again for aiding me in this.

Cheers,

Veera   

[veera jaime]

Dear Alexei,

Thank you so much for giving concern over my predicaments here. Indeed, BEAST2 is a great package with more user friendly features in various ways especially when it comes to xml file preparation via GUI in BEAUti. Please find attached the log files from two separate runs namely mdataset5disBEAST and mdataset5disBEAST2, I've also attached the log combined file of these two log files (mdataset5disBEASTcombo) without any burn-in settings. I'm so glad that this could be fixed, thank you so much again for aiding me in this.

Cheers,

Veera   

On Friday, February 15, 2013 2:54:19 AM UTC+8, Alexei Drummond wrote:

[veera jaime]

Hi Remco,

Thank so much you for the suggestions. I've tried to edit the log file according to the suggestions below. I managed to load the edited log file in SPREAD, neverthless, unable to generate the kml file. The program terminated with error message " Unexpected problem on thread SwingWorker pool-2-thread-2: Index:0 Size:0 null". Do let me know if I need to further edit the log file (or if I've made errors in the editing). Please find attached the edited log file. I've also attached the original log files in the previous threads. Thank you again for assisting me on this.

Cheers,

Veera

[Remco Bouckaert]

Hi Veera,

I am not sure what causes the error message, but I think that indicators
should be all lower-case. The log file you sent had names as
'Indicators' instead of 'indicators'. With this change and some dummy
locations I managed to get a kml output file. Work is under way to come
up with a more permanent fixed so no more log file editing is required.

Cheers,

Remco

[veera jaime]

Hi Remco,

Thanks for the update. I will do the changes to the log file and try to run it again. Hopefully, could get it through this time around. Will keep you updated then.

Cheers,

Veera

[Mohammed Houta]

Dears Dr. Remco,

Greetings,

I followed the following instructions:

A quick fix to do BFs in SPREAD is to edit the log file as follows:

1. remove all comments -- those are lines starting with a hash -- from
the start of the file

2. rename the rateIndicator.location columns to indicator. This is
assuming you called your partition 'location', if it is something else,
there should be columns named rateIndicator.<partition name>. The
columns should be numbered now and called indicator1, indicator2,
indicator3, etc.

 but, unfortunately,  I get the same error, {no indicators columns found. i suspect wrong or malforming log file}

Thanks in advance,

M. Houta