credibleSet only producing one tree

24 views
Skip to first unread message

Mael Glon

unread,
Oct 5, 2020, 9:29:10 AM10/5/20
to ExaBayes
Hello, 

I have now used Exabayes on two anchored hybrid enrichment datasets. Each run ran for a few million generations and all ESS were >200, and I have no issues summarizing the runs with consense, postProcParam, and extraBips. I was hoping to use Densitree to look at the distribution of trees so I used the credibleSet command as follows:

credibleSet -f ExaBayes_topologies.* -n cred

After a few seconds, the following message was printed in the terminal:

printed the 50% credible set to ExaBayes_credibleSet.cred

However, when I looked at the resulting file, there is only one tree. Here is the full text from that file:

freq tree

12354 (678_Glon,669_Glon,((209_Glon,63_Glon),((((571_Glon,453_Glon),(594_Glon,(402_Glon,403_Glon))),((1_Glon,(488_Glon,484_Glon)),(529_Glon,(153_Glon,528_Glon)))),((231_Glon,(175_Glon,26_Glon)),(04_Crandall,(((664_Glon,193_Glon),((399_Glon,(601_Glon,(597_Glon,596_Glon))),((659_Glon,(423_Glon,422_Glon)),(164_Glon,(523_Glon,(610_Glon,611_Glon)))))),(((438_Glon,(506_Glon,502_Glon)),(166_Glon,(11_Glon,435_Glon))),((204_Glon,(169_Glon,521_Glon)),((345_Glon,(344_Glon,655_Glon)),((134_Glon,132_Glon),(31_Glon,(142_Glon,141_Glon))))))))))));

My understanding is that credibleSet should provide quite a few more trees (I'm using the 50% default setting). 

I would appreciate any insight into what I may be doing wrong and how to fix this.

Thanks!

Alexandros Stamatakis

unread,
Oct 6, 2020, 4:10:46 AM10/6/20
to exab...@googlegroups.com
Hi Mael,

I'd guess that the * in "credibleSet -f ExaBayes_topologies.* -n cred"

might generate the problem.

What happens if you pass a file name without a wildcard to credibleSet ?

Alexis

On 05.10.20 16:29, Mael Glon wrote:
> Hello,
>
> I have now used Exabayes on two anchored hybrid enrichment datasets.
> Each run ran for a few million generations and all ESS were >200, and I
> have no issues summarizing the runs with consense, postProcParam, and
> extraBips. I was hoping to use Densitree to look at the distribution of
> trees so I used the credibleSet command as follows:
>
> *credibleSet -f ExaBayes_topologies.* -n cred*
>
> After a few seconds, the following message was printed in the terminal:
>
> *printed the 50% credible set to ExaBayes_credibleSet.cred*
>
> However, when I looked at the resulting file, there is only one tree.
> Here is the full text from that file:
>
> *freqtree*
>
> *12354(678_Glon,669_Glon,((209_Glon,63_Glon),((((571_Glon,453_Glon),(594_Glon,(402_Glon,403_Glon))),((1_Glon,(488_Glon,484_Glon)),(529_Glon,(153_Glon,528_Glon)))),((231_Glon,(175_Glon,26_Glon)),(04_Crandall,(((664_Glon,193_Glon),((399_Glon,(601_Glon,(597_Glon,596_Glon))),((659_Glon,(423_Glon,422_Glon)),(164_Glon,(523_Glon,(610_Glon,611_Glon)))))),(((438_Glon,(506_Glon,502_Glon)),(166_Glon,(11_Glon,435_Glon))),((204_Glon,(169_Glon,521_Glon)),((345_Glon,(344_Glon,655_Glon)),((134_Glon,132_Glon),(31_Glon,(142_Glon,141_Glon))))))))))));*
>
> My understanding is that credibleSet should provide quite a few more
> trees (I'm using the 50% default setting).
>
> I would appreciate any insight into what I may be doing wrong and how to
> fix this.
>
> Thanks!
>
> --
> You received this message because you are subscribed to the Google
> Groups "ExaBayes" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to exabayes+u...@googlegroups.com
> <mailto:exabayes+u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/exabayes/ea94038e-ee4e-4ded-9e86-391e31016eean%40googlegroups.com
> <https://groups.google.com/d/msgid/exabayes/ea94038e-ee4e-4ded-9e86-391e31016eean%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

Mael Glon

unread,
Oct 6, 2020, 8:43:18 AM10/6/20
to ExaBayes
Hi Alexis,

Thanks for your reply. I tried your suggestion but I still only got one tree in the credible set file.

Thanks,
Mael

Andre Aberer

unread,
Oct 6, 2020, 10:32:07 AM10/6/20
to exab...@googlegroups.com
Hi Mael,

if this particular tree occurs more than half of the time in the
'topology file(s)' containing the sampled trees, then the 50% credible
set of trees consists of precisely this topology. If there is strong
signal in the data, this is not impossible. Did you have a look at the
sampled trees?

What about a higher percentage credible set? E.g., for 75%, use
parameter "-c 75". With -c 100 you should effectively get all
sampled topologies ranked by frequency of occurrence (I do not
accurately remember the output format though).

That being said, if you executed a single run only and the sampled trees
mostly have the same topology, then there is a chance that the chain has
gotten stuck and is not representative. With Bayesian MCMC it is
important to verify that multiple distinct runs yield the same result.

--
Best regards,
Andre

Mael Glon

unread,
Oct 6, 2020, 10:42:11 AM10/6/20
to ExaBayes
Hi Andre,

Thanks for the suggestion. As I increase the percentage of the credible set I do get more trees. I did start with parsimony trees rather than random trees, and I expect this dataset to have very strong signal because of my sampling depth and breadth, so perhaps this is normal.

Thanks,
Mael

Alexandros Stamatakis

unread,
Oct 7, 2020, 1:38:24 AM10/7/20
to exab...@googlegroups.com
If you want to be absolutely sure, you can compute the pair-wise
distances among all trees in your tree set with the corresponding RAxML
or RAxML-NG option, this should also report you the number of unique
trees in the tree set.

Alexis

PS: Andre, many thanks for all your help :-)
> >> www.exelixis-lab.org <http://www.exelixis-lab.org>
> >>
>
> --
> You received this message because you are subscribed to the Google
> Groups "ExaBayes" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to exabayes+u...@googlegroups.com
> <mailto:exabayes+u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/exabayes/e05366b6-ff04-48fb-b2d0-725faf82eb6dn%40googlegroups.com
> <https://groups.google.com/d/msgid/exabayes/e05366b6-ff04-48fb-b2d0-725faf82eb6dn%40googlegroups.com?utm_medium=email&utm_source=footer>.

Alexandros Stamatakis

unread,
Oct 7, 2020, 1:39:19 AM10/7/20
to exab...@googlegroups.com
I mean to write: pair-wise RF (Robinson Foulds) topological distances
among all pairs of trees in your tree set,

Alexis

On 06.10.20 17:42, Mael Glon wrote:
> >> www.exelixis-lab.org <http://www.exelixis-lab.org>
> >>
>
> --
> You received this message because you are subscribed to the Google
> Groups "ExaBayes" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to exabayes+u...@googlegroups.com
> <mailto:exabayes+u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/exabayes/e05366b6-ff04-48fb-b2d0-725faf82eb6dn%40googlegroups.com
> <https://groups.google.com/d/msgid/exabayes/e05366b6-ff04-48fb-b2d0-725faf82eb6dn%40googlegroups.com?utm_medium=email&utm_source=footer>.

Mael Glon

unread,
Oct 7, 2020, 8:40:22 AM10/7/20
to ExaBayes
Great, thank you both for your help!
Reply all
Reply to author
Forward
0 new messages