Error with uniform priors and fixed topology

[Edgardo Ortiz Valencia]

Dear beast-users:

I am getting the following error in v2.4.4 when using 4 node calibrations with uniform priors and a fixed topology. However, the file runs fine when I remove the fixed tree or when I use fixed tree with normal priors for the node calibrations:

Start likelihood: -Infinity after 10 initialisation attempts
P(posterior) = -Infinity (was -Infinity)
    P(prior) = -Infinity (was -Infinity)
        P(YuleModel.t:treeV) = -3.5150001669535413 (was -3.5150001669535413)
        P(YuleBirthRatePrior.t:treeV) = 0.0 (was 0.0)
        P(GammaShapePrior.s:ITS1_5_8S_ITS2) = -1.0 (was -1.0)
        P(GammaShapePrior.s:matK) = -1.0 (was -1.0)
        P(GammaShapePrior.s:ndhF) = -1.0 (was -1.0)
        P(KappaPrior.s:ITS1_5_8S_ITS2) = -1.8653600339742873 (was -1.8653600339742873)
        P(PropInvariantPrior.s:matK) = 0.0 (was 0.0)
        P(PropInvariantPrior.s:ndhF) = 0.0 (was 0.0)
        P(RateACPrior.s:matK) = -3.184008455701433 (was -3.184008455701433)
        P(RateACPrior.s:ndhF) = -3.184008455701433 (was -3.184008455701433)
        P(RateAGPrior.s:matK) = -6.915086640662835 (was -6.915086640662835)
        P(RateAGPrior.s:ndhF) = -3.1686658147294304 (was -3.1686658147294304)
        P(RateATPrior.s:matK) = -6.915086640662835 (was -6.915086640662835)
        P(RateATPrior.s:ndhF) = -3.184008455701433 (was -3.184008455701433)
        P(RateCGPrior.s:matK) = -6.915086640662835 (was -6.9150866406Fatal exception: Could not find a proper state to initialise. Perhaps try another seed.
62835)
        P(RateCGPrior.s:ndhF) = -3.184008455701433 (was -3.184008455701433)
        P(RateGTPrior.s:matK) = -6.915086640662835 (was -6.915086640662835)
        P(RateGTPrior.s:ndhF) = -6.915086640662835 (was -6.915086640662835)
        P(MeanRatePrior.c:clockV) = -2.402585092994046 (was -2.402585092994046)
        P(ucldStdevPrior.c:clockV) = 0.819364145120802 (was 0.819364145120802)
        P(crownL.prior) = -Infinity (was -Infinity)
        P(crownV.prior) = NaN (was NaN)  **
        P(stemA.prior) = NaN (was NaN)  **
        P(stemC.prior) = NaN (was NaN)  **
    P(likelihood) = NaN (was NaN)  **
        P(treeLikelihood.ITS1_5_8S_ITS2) = NaN (was NaN)  **
        P(treeLikelihood.ndhF) = NaN (was NaN)  **
        P(treeLikelihood.matK) = NaN (was NaN)  **
java.lang.RuntimeException: Could not find a proper state to initialise. Perhaps try another seed.
    at beast.core.MCMC.run(Unknown Source)
    at beast.app.BeastMCMC.run(Unknown Source)
    at beast.app.beastapp.BeastMain.<init>(Unknown Source)
    at beast.app.beastapp.BeastMain.main(Unknown Source)
    at beast.app.beastapp.BeastLauncher.main(Unknown Source)
Fatal exception: Could not find a proper state to initialise. Perhaps try another seed.

BEAST has terminated with an error. Please select QUIT from the menu.

Thanks, any advice is appreciated


Edgardo

[Remco Bouckaert]

Hi Edgardo,

The error message indicates that the prior "crownL.prior“ is -Infinity, so the initial tree is violating that constraint.

Perhaps, the fixed tree does not conform to that constraint?

If it does conform, when you add a fixed tree, did you remove the random tree used to initialise the tree?

Cheers,

Remco

--
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.

[Edgardo Ortiz Valencia]

Hi Remco,

Yes, I removed the random tree (following instructions in "All about starting trees") also set the corresponding operators to 0. I checked and it doesn't violate the constraints, my fixed tree is just a topology without branchlengths. Also, the file runs OK when using normal distributions for the priors with the same fixed topology.

[Remco Bouckaert]

Hi Edgardo,

Thanks for the XML file — it looks like there is no timing information in the starting tree (no branch lengths in the Newick tree), and the calibrations have hard boundaries. After the tree is scaled (so that tips all have age zero), these boundaries are violated resulting in -Infinity log posterior at the start.

Do you have a timed tree that you can use as starting tree that does not violate the calibrations?

Cheers,

Remco

[Edgardo Ortiz Valencia]

Thanks! I have the same tree with branch lengths from a ML analysis (not time-scaled), I could also use the tree I will get from the analysis with normal priors. What would be the best option?

Edgardo

[Remco Bouckaert]

I suppose the tree with the topology that you want to check is the one to use.

Branch lengths will be estimated, so if both topologies are the same it does not matter which one you choose, as long as they fit the constraints.

Remco