Restarting problem in MrBayes Checkpoint .

[Debajyoti Kabiraj]

Dear experts,

 I've run MrBayes in our institute HPC. In first I have run 10 million generations and I got an average standard deviation of split frequencies: 0.135901. Then again I have run for another 10 million generations using append=yes and got average standard deviation of split frequencies: 0.145050. After that when I'm trying to run another 10 million generation then I got below error.

 1. Use the same sampling frequency as in the previous run to use relative burnin.

      2. Check (and modify) the number in [generation: number] at line 3 of the .ckp file

         to match the previous number of generations in all the .p and .t files. This may

         happen if checkfreq was smaller than samplefreq.

      3. Rarely, delete the last sample/line in the .p and .t files to achieve 2. above.

         This may happen if ngen was not divisible by samplefreq.

      Error preparing print files

      Error in command "Mcmc"

      The error occurred when reading char. 6 on line 5

         in the file 'Combined_gbP_ctl_3.nex'

   Error in command "Execute"


Then I'm trying to follow this tutorial 
http://www.phylo.org/index.php/tools/mrbayes_321restartbeta.html

But the problem is that the last generation of  .p and .t  files are different and lesser than .ckp.
I have attached all the files kindly guide me on what to do.

[Mark Miller]

Hello Debajyoti, 

Thanks for reporting the issue. This is not an issue I have encountered before, so I am just guessing.

I would save copies of the relevant files. and edit the copies as recommended in  step2:

 

    2. Check (and modify) the number in [generation: number] at line 3 of the .ckp file

         to match the previous number of generations in all the .p and .t files. This may

         happen if checkfreq was smaller than samplefreq

I would delete lines as needed in .p and .t files so the number of generations are  equal in both, and then match those to the ckp file.

Best,

Mark