search

pgfoster / p4-phylogenetics

A Python phyloinformatic toolkit, and an implementation of tree-heterogeneous models of evolution
GNU General Public License v2.0
28 stars 8 forks source link

MCMC Aborted error! #16

Open 09Raghu09 opened 2 years ago

09Raghu09 commented 2 years ago

Dear Sir,

I tried to run Bayesian analysis with MCMC using p4. After certain time period I get this error everytime. I used the bacterial16s data. Could you please check this and if you solve this issue, that would be really helpful.

However, I tried following your you tube video regarding the MCMC explanation. I want to ask, after successful running p4, the output we get is the collection of maximum likelihoods or the best tree. If it is the collection, how can I get the best tree out of it. If you give some insights that would be great

................................................. 450000 - 00:07:57 ................................................. Mcmc.checkPoint(), chainNum 0. Bad likelihood calculation just before writing the checkpoint. Old curTree.logLike -4556.17, new curTree.logLike -5423.71, diff 867.538 500000 - 00:07:12 ................................................. 550000 - 00:06:23 ................................................. 600000 - 00:05:42 ................................................. 650000 - 00:04:59 ................................................. 700000 - 00:04:36 .................................................corrupted size vs. prev_size /usr/bin/python3: line 2: 32012 Aborted (core dumped) prun python3 "python3" "$@"

Best

pgfoster commented 2 years ago

Dear 09Raghu09,

It is not obvious to me what the problem is. Perhaps you can send me the dataset and script that you used to run it? That might help me to debug it. You mentioned a youtube video --- which video is that? MCMC samples are not ML; It is a bit hard to explain the process in a few words. best, Peter

09Raghu09 commented 2 years ago

Dear @pgfoster

I used the same script which is there in the documentation of p4 with the headline " A simple MCMC" under Bayesian analysis with MCMC (link attached for reference). However, for the sequence, For the dataset, I took bacterial16s in Fasta format (you used d.nxs), and for the random tree, I used the tree below.

(Thermotoga:0.055774,(Thermus:0.061273,(Bacillus:0.102121,Deinococcus:0.106049):0.044776):0.040848,Aquifex:0.110762);

https://p4.nhm.ac.uk/tutorial/tut_mcmc.html

No problem, I just want to know, what will be the output look like after the execution. I think it will be the series of likelihood values. If so, how exactly to get the best tree to plot.

pgfoster commented 2 years ago

@09Raghu09, Sorry, the documentation is out of date (I hope to fix that soon ...). The examples in /share/Examples are more up-to-date. As for your questions about results from an MCMC analysis, that would need a more in-depth answer than I can give here.