New version of StarBEAST2 (v0.13)

[Alexei Drummond]

Dear StarBEAST2 users,

We have released version 0.13 of StarBEAST2 with further improvements to computational performance. It can be updated on your system through the package manager of the latest version of BEAUti (BEAST 2.4.3). 

Beyond the original improvements laid out in the manuscript, we have now also optimised the calculation of coalescent times within each species branch* and developed an optimised version of the NodeReheight operator - the most often used operator in StarBEAST2.

StarBEAST2 v0.13 is much faster than both *BEAST and previous releases StarBEAST2. We are working on quantifying the exact impact on performance, and will update the StarBEAST2 manuscript to reflect the performance of this version.**

Unfortunately, this update is not compatible with previously generated XML files. So to take advantage of this new version, update StarBEAST2 to version 0.13 using the package manager, then generate new XML files from scratch using BEAUTi.

Regards,

Huw Ogilvie, Remco Bouckaert and Alexei Drummond

* These optimisations include replacing recursion with iteration, and using primitive arrays instead of computationally-expensive classes.

** The manuscript currently available on bioRxiv (http://www.biorxiv.org/content/early/2016/08/18/070169) is based on an earlier version of StarBEAST2.

[Mark Miller]

Dear Alexei et al.

We are trying to install StarBEAST2 from GitHub, on our Linux system but there does not seem to be a .jar file attached to this package.
Can you explain how we should proceed?

Best,

Mark Miller

[Tim Vaughan]

Hi Mark,

The best way to install the package is via BEAUti's package manager (File->Manage Packages).  On a GNU/Linux system you can also use the command line addonmanager tool.  If neither of these are appropriate (perhaps because they require network access) you can download the ZIP file containing the compiled latest release from https://github.com/genomescale/starbeast2/releases/latest (currently this is StarBeast2.zip).  You'll need to create a directory in ~/.beast/2.4/ named StarBEAST2 and unzip the archive in that directory.

Hope this helps,

Tim

--
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 https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

[Miller, Mark]

Thanks Tim,

The link I had led to a repo that was missing the jar file.

Now we are ready to roll…

Mark

[EvoClive]

Hi Everyone

I notice that under the new StarBEAST2 v0.13 release that the File>Template tab now gives the extra option "StarBeast" as opposed to the "SpeciesTreeRLC" option I was using under the most recent manual I found. Additionally, there are different tabs (e.g. "multi species coalescent") under this new template. Does this change much when setting up species tree analyses or is there any user manual for this new implementation?

Many thanks

Clive

[Huw A. Ogilvie]

Hi Clive, the "StarBeast" template is for strict clock analyses, and the "SpeciesTreeXXX" templates are for strict clock analyses. This is explained in the tutorials available on GitHub - https://github.com/genomescale/starbeast2/releases

[Pablo G Goikoetxea]

Dear developers and users

I have started analyzing my data with the brand new StartBeast2 using Beast 2.4,5. My data consists of sequence data (6 sequences) from two species, sampled at several locations. Analyses for 3 localities run smoothly (for example,  SB2_Fer.xml), but I have two localities with the recurrent error "Fatal exception: Could not find a proper state to initialise. Perhaps try another seed" (attachments SB2_Sot.xml and SB2_Tal.xml).

I guess some of the priors might be wrong, but I am not sure where to start looking at. I have read previous posts on the same issue, but could not find an appropriate answer (I think). I should appreciate any help.

Pablo

[Huw A. Ogilvie]

Hi Pablo, for some reason the default initialization method is failing with your data.

To solve this issue, simple change the init element, which looks like this:

<init id="SBI" spec="starbeast2.StarBeastInitializer" birthRate="@speciationRate.t:Species" estimate="false" speciesTree="@Tree.t:Species">

to use random initialization, by adding method="random" to it, like so:

<init id="SBI" spec="starbeast2.StarBeastInitializer" birthRate="@speciationRate.t:Species" estimate="false" speciesTree="@Tree.t:Species" method="random">

[Pablo G Goikoetxea]

Thanks a lot Huw!! Working like a charm.

Cheers

Pablo

--
You received this message because you are subscribed to a topic in the Google Groups "beast-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/beast-users/Q3z_RghkoUo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to beast-users+unsubscribe@googlegroups.com.

To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

--

-------
Pablo G Goikoetxea
Don Vela 54, 2º izda
01009 Vitoria-Gasteiz (SPAIN)

[Pablo G Goikoetxea]

Dear Huw

I have finished comparing my 2 sp gene trees through different locations (thanks again for your help, Huw!!) and now I am trying to set up a different type of analysis in StarBeast2.

I am using one species, sampled at five localities (8-12 haploid data per locality) and 6 re-sequenced genes. My objective is to use the localities as OTUs (i.e., populations) to obtain a multiple marker tree with a calibrated root (following the example in Tutorial 2). I expect this would give me an idea of intra-specific evolution.

The priors tab in Beauti shows my 5 populations as monophyletic groups, by default. I have kept them monophyletic and selected Exponential priors with estimated means. The root.t:prior has been established as instructed in the tutorial, monophyletic with mean 0.9 and offset 1.5 (yes, a large calibration range...).

Unfortunately, I have run into an "Error 110 Taxon not found in tree" (see attachment). Looking into the Beast googlegroups I have found some similar problems caused by Beauti adding some 1 to the names. However, this does not seem to be my case.

Furthermore, I would like to know whether the settings for the monophyletic populations are appropriate and whether the hyperpriors should be left as they are (not commented in the Tutorial).

Thanks for any help. I am attaching the xml file too.

Pablo

On Mon, Apr 24, 2017 at 7:08 AM, Huw A. Ogilvie <auss...@gmail.com> wrote:

--

You received this message because you are subscribed to a topic in the Google Groups "beast-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/beast-users/Q3z_RghkoUo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to beast-users+unsubscribe@googlegroups.com.
To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

[Nelson]

Dear Pablo, please let me ask a question

If I got it right, you're doing a population (location) tree for one of your species using the starbeast approach to account for genealogical discordance among gene trees, right? I'm not sure this will give you accurate divergence times unless you can assume that there is virtually no gene flow among locations (populations, your OTUs), as any genealogical discordance caused by gene flow will be modelled as genealogical discordance due to incomplete lineage sorting. I'd guess that your analysis may underestimate divergence times if there is significant gene flow among locations. Am I right or am I missing something here?

My question is, does anyone know how robust is the starbeast model against violations caused by gene flow? For example, in cases of speciation with gene flow, how much gene flow affects diverge time estimates (if it does)?

Cheers
Nelson

To unsubscribe from this group and all its topics, send an email to beast-users...@googlegroups.com.

To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

[Alexei Drummond]

Nelson,

This paper may have some partial answers to your question:

https://bmcevolbiol.biomedcentral.com/articles/10.1186/1471-2148-13-44

Cheers

Alexei

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.

[Graham]

Nelson,

This article looks at the problem in detail.
https://academic.oup.com/sysbio/article/63/1/17/1687129/The-Influence-of-Gene-Flow-on-Species-Tree

I am working on a BEAST2 package DENIM for species tree estimation in the presence of small amounts of migration, see http://indriid.com/workingnotes2017.html. I don't think it will help much with gradual speciation, but when there is even a little migration between non-sister species, it is very disruptive to *BEAST, and DENIM is more robust to that.

Graham Jones

[Pablo G Goikoetxea]

Dear Nelson

Yes, you are right!!. And I am assuming little or no gene flow among populations (based on SSRs and geographic data).

I think you are touching the core of my (ultimate) question to Beast: can I separate incomplete lineage sorting from inter-specific migration in the 2 species from each locality?. 

My previous analysis (gene trees from each locality, the two species together)

suggests several candidates for inter-specific migration or admixture after secondary contact (attachment). But, is it enough? I recognize I'm fishing here, with the new analysis I intended to make, just trying to gain some more evidence(s)/support.

Any advise is absolutely welcome.

Thanks Alexei and Graham for the excellent literature.

Cheers

Pablo

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

To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

[Nelson Fagundes]

Alexei and Graham, thanks for the papers!

Pablo, perhaps you could use IM (IMa2?) to estimate migration parameters and check how large they are. However, I'm not sure if the IM output would allow you to formally test if a scenario with migration is better than a scenario with full isolation (but I'm sure you can turn migration off).

Graham, I'm not sure if it would be feasible, but would certainly be nice to have an option to restrict migration during recent times (up to the user?) to try to discrminate between secondary contact or a more continuous gene flow scenario. I'm looking foward to play a bit with the DENIM package!

Cheers

Nelson

--

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+unsubscribe@googlegroups.com.

[Pablo G Goikoetxea]

Thank you Nelson.

I am not sure I should use the IM family of programs, at least for intra-specific analysis, as my data surely violate the main assumption (that there is no other populations more closely related to the sampled ones than they are to each other).

However, it might be worth trying IM with some of the inter-specific population pairs (they are marginal populations at the limits of both species distribution ranges).

I strongly agree with your suggestion for DENIM (if possible, indeed), although I think the options should be the continuous migration scenario vs no-migration->secondary contact->no migration or even scenario 1 versus no-migration -> secondary contact -> migration

Cheers

Pablo

[Pablo G Goikoetxea]

Sorry, Nelson & Graham.

My last sentence regarding DENIM scenarios was nosense... Any options, if available, should be valuable and very much welcome!!

Thanks for sharing

Pablo

On Tue, Apr 25, 2017 at 6:17 PM, Nelson Fagundes <nros...@gmail.com> wrote:

[Master Cheng]

Hi 

I want to know how to resume analyses in BEAST2,i did not see the related tutorials.

Thanks.

Jinjin

在 2016年10月10日星期一 UTC+8上午8:31:46，Alexei Drummond写道：

[Master Cheng]

Dear Tim

    I want to know how to resume the analyze in Beast2,Could you sent me the relation the tutorial？

Thanks 

Jinjin

在 2017年1月27日星期五 UTC+8上午4:38:42，Tim Vaughan写道：

To unsubscribe from this group and stop receiving emails from it, send an email to beast...@googlegroups.com.