strange calibration

[pasca...@gmail.com]

Hi all,

I'm setting up an mcmctree analysis. This is a 10-partition dataset for a 364 tip tree and 8 node calibrations.

I prepared the node calibration information using the MCMCtreeR package.

When I run mcmctree (v4.10.7), it lists details about the different partitions, then outputs the following:

Fossil calibration information used.
Node 365:   B (  8.2360, 10.0750,  0.0100,  0.0500 )
Node 367:   B (  6.1600,  8.3800,  0.0100,  0.0500 )
Node 368:  SN (  2.0000,  1.0200,  7.0000 )
Node 372:  SN (  1.0500,  1.0200,  7.0000 )
Node 575:  SN (  1.2000,  1.0200,  7.0000 )
Node 623:  SN (  0.3500,  1.0200,  7.0000 )
Node 657:  SN (  0.9000,  1.0200,  7.0000 )
Node 723:   B (  3.5320,  6.4520,  0.0100,  0.0500 )
Node 724:   B (  8.2360, 10.0750,  0.0100,  0.0500 )
Node 726:   B (  6.1600,  8.3800,  0.0100,  0.0500 )
Node 727:  SN (  2.0000,  1.0200,  7.0000 )

500000 burnin, sampled every 500, 20000 samples
Approximating prior
(Settings: cleandata=0  print=2  saveconP=0)

getting initial values to start MCMC.


daughter node 724 minimun age    8.2360
mother node 723 max age    6.4520

error: strange calibration?

I thought this might mean that there is a conflict with the node calibrations, but I don't see it... From examining the tree, node 723 does in fact have a max age specified at 6.452, but node 724 should NOT have a minimum age of 8.236. That value belongs to node 365, as is shown in the printed table.

You can download the input files here if you would like to replicate.

Any thoughts?

Thanks!

-Pascal

[pasca...@gmail.com]

Actually, I see the problem in FigTree. I just need to determine how this is happening. I will report back.

[pasca...@gmail.com]

This may have been a bug in the MCMCtreeR package. I was able to adjust the code in the R package and that solved this problem.

[Sandra AC]

Hi there!

Please note that the MCMCtreeR R package is independent from PAML, and hence you may have to contact its developer separately to deal with this potential bug (e.g., you could raise an issue in their GitHub repository where you explain the problem you are experiencing and provide the developer with as many details as possible to help them to troubleshoot this error). To learn more about the notation used to write the calibrations compatible with MCMCtree, you should check the PAML documentation (pp. 49-52).

Meanwhile the issue with the MCMCtreeR R package is sorted, you should always compare the notation that this R package outputs to the expected MCMCtree notation as detailed in  the PAML documentation. If you want to visualise lower-bound, upper-bound, and soft-bound calibration densities, you can use the mcmc3r R package (an explanation is given in the `README.md` file). To plot SN and ST distributions, you can use the R package `sn` (i.e., `sn::dsn`, `sn::dst`) if you know the parameter values of such distributions.

Hope this helps!

S.

[pasca...@gmail.com]

Yes, understood - thank you!

I was just briefly reporting back here, in case others found themselves in a similar situation. Once I confirm what I think is going on, I will post an issue to the MCMCtreeR github.