Migrate-N

The software and instructions can be found on: ​http://popgen.sc.fsu.edu/Migrate/Info.html

Mykopat Grid specifics

Since Migrate-N can utilise multiple cores/computers when calculating migration rates, there are some specifics to keep in mind when running it in parallel. The basic workflow will be something like this (completely untested by me/MBD):

1. Start migrate-n by typing migrate-n my_params.txt
2. Using the menus in migrate and the manual, set all parameters for your run, but DO NOT start the run.
3. Save the param file, theres a menu option for that.
4. Create a batch file used to submit the job to the grid. An example can be found at the bottom of the page. Request a small amount of memory per slot (1GB should be enough), but enough runtime for finishing the run.
5. Submit the job with qsub my_batch_file and wait for an email telling you first that the job started to run, and then yet another when the run is done.
6. Check the log files (job_name.oNNN and job_name.eNNN) to ensure that everything went well.
7. If you need to rerun, start over from the top of this list, but the batch file can be reused.

#!/bin/sh
# Sample batch file for migrate. Put this in a plain text file called, for example migrate_batch and call it with qsub migrate_batch

# Ensure the environment is correct
#$ -V
#$ -cwd

# Ask for a mail at beginning, end and if the job is aborted
#$ -m bea

# We want to run migrate on 6 slots/cores, this number should ideally be number of loci + 1. 
# However, there is currently a system limit at 8. The more you ask for, the longer it usually
# takes before your job is started.
#$ -pe openmpi 6

# We want the system to allocate 5 hours run time to migrate, and 1G of ram per slot
#$ -l h_rt=5:0:0,h_vmem=1G


# Change my_parm_file to the name of your param_file
mpirun migrate-n-mpi my_parm_file -nomenu