Help with ClaDS Setup in BEAST2 (v2.7.8)

[Alfredo Sousa]

Hi everyone,

I’m working on estimating species diversification rates using ClaDS in BEAST2 (v2.7.8) and would appreciate any advice on my setup.

I’ve been following the ClaDS tutorial (https://taming-the-beast.org/tutorials/ClaDS-tutorial/) and generated the XML file using BEAUti. Here's a summary of my configuration:

Template: ClaDS (selected in BEAUti)

Alignment: Amino acid sequences, 47 taxa, ~2200 sites

Site model: OBAMA with multiple models selected

Clock model: Relaxed Clock Log Normal

Calibration: MRCA prior on a monophyletic clade, using a LogNormal distribution (M = 6.787, S = 0.09; centered ~600 Ma)

Tree prior: ClaDS Birth-Death with ρ = 0.3

ucldMean.c: Exponential prior with mean = 0.03

Other settings: Defaults unless stated; chain length = 10 million, pre-burnin = 1 million

The analysis fails to initialize, with P(posterior) = -Infinity, treeLikelihood = NaN, and the MRCA prior flagged as -Infinity. Please, find the log attached. I suspect the issue may relate to the MRCA calibration or the input alignment. I’ve tried adjusting parameters and using different gene subsets (my data comes from a larger phylogenomic matrix of ~220 genes / ~70,000 amino acid sites), but without success.

Does anyone have suggestions on how to improve or troubleshoot this setup?

Many thanks in advance,
Alfredo

[Joëlle Barido-Sottani]

Hello Alfredo,

Have you set a starting tree in your analysis, and if so can you make sure that the ages and topology of that starting tree are compatible with your MRCA prior ?

Otherwise this may be a bug in ClaDS - could you send me your XML so I can do some troubleshooting ?

Best,

Joëlle