launch_single_cosi_iteration.sh

#!/usr/bin/env bash
# Wrapper to launch a single cosi iteration
#
# COSI ( http://www.broadinstitute.org/~sfs/cosi/ ) is a software to simulate Human Genetic Variation
# (see http://genome.cshlp.org/content/15/11/1576.full for the reference).
# It can be used to simulate SNP data of 3 human populations, EUR, AFR and ASN.
#
# This script is a wrapper to launch multiple COSI simulations in parallel, using the GNU/parallel tool
#
#
#
# INSTALLMENT AND REQUIRED FILES
#
# 1. Download the cosi tool from http://www.broadinstitute.org/~sfs/cosi/
# 2. create a new folder, e.g. mkdir cosi_run
# 3. copy this script in the new folder
# 4. copy the following files from the cosi .tar distribution to the new folder: 
#	cosi/coalescent 
#	cosi/recosim
# 	examples/bestfit/autosomes_deCODE.distr
#	examples/bestfit/model.test
#	examples/bestfit/params
#	examples/bestfit/recParams
#    (alternatively, use your own params files)
#
# 5. Launch the simulations, using GNU/parallel:
#
# 	$: seq 1 100 | parallel "./launch_single_cosi_iteration.sh {} outputfolder"
#
#    See below for more options.
#
# 
#
#
# USING THIS SCRIPT TO LAUNCH COSI SIMULATIONS
#
# 
# There are two recommended ways to run this script: from bash using parallel, or from 
# a sun grid engine environment using qsub. To use one or another, you need to uncomment the lines
# defining REPLICA_ID and OUTPUTDIR.
#
# - Launching from bash using parallel:
#
#    Make sure you are using GNU/parallel and not the standard parallel from moreutils.
#    The syntax should be only slightly different between the two:
# 
#       Gnu/Parallel:
#
#           $: seq 1 100 | parallel "./launch_single_cosi_iteration.sh {} outputfolder"
#
#       Parallel from moreutils:
#
#           $: seq 1 100 | parallel -i sh ./launch_single_cosi_iteration.sh {} outputfolder
#
#
#
#
# - Launching in a Sun Grid Engine environment:
# 
#    Customize the qsub options below (any line beginning with #$ will be interpreted as a qsub parameter).
#
#    Then, launch the script from qsub:
#
#           $: qsub launch_single_cosi_iteration.sh
#
# QSUB options
#
#$ -cwd
#$ -l h_vmem=600M
#$ -j y
#$ -N cosi5000
#$ -o logs
#$ -V
# Use the -t option to define how many simulations you want.
# E.g. -t 1-1000 will launch 1000 simulations
#$ -t 3-1000
#
# 
# NOTES TO THE CODE IN THIS SCRIPT:
#
# * Each iteration is run in a single directory, defined by $ITERATION_ID 
# * All the files necessary are copied in the directory
# * The recombination model is calculated by recosim for each iteration.
# 
#   $1 -> iteration number (usually I execute 100 iteration for each set of parameters)
#   $2 -> base output folder
# 

##### CONFIGURATION

### Parallel - uncomment the next two lines if you are using parallel:
REPLICA_ID=$1
OUTPUTDIR=$2

### SGE environment - uncomment the next two lines if you are using SGE:
#REPLICA_ID=$SGE_TASK_ID
#OUTPUTDIR=../../data/simulations_5000

# Copy all cosi binaries to a temporary subfolder of the output directory
mkdir -p ${OUTPUTDIR}/tmp/${REPLICA_ID}
mkdir -p ${OUTPUTDIR}/param_logs
mkdir -p ${OUTPUTDIR}/results
cp autosomes_deCODE.distr params recParams recosim coalescent ${OUTPUTDIR}/tmp/${REPLICA_ID}
#cp autosomes_deCODE.distr recosim coalescent ${OUTPUTDIR}/${REPLICA_ID}
cp recParams ${OUTPUTDIR}/tmp/${REPLICA_ID}/bestfit_neutral_replica_${REPLICA_ID}
cd ${OUTPUTDIR}/tmp/${REPLICA_ID}

# generate the recombination model (model.test), and run cosi
./recosim bestfit_neutral_replica_${REPLICA_ID} 1000000
./coalescent -p ./params -o ../../results/bestfit_neutral_replica_${REPLICA_ID}
gzip -f ../../results/bestfit_neutral_replica_${REPLICA_ID}.*

# Clean the temporary subfolder, and move all results to output dir
rm coalescent recosim
mv model.test ../../param_logs/bestfit_neutral_replica_rec_${REPLICA_ID}
gzip -f ../../param_logs/bestfit_neutral_replica_rec_${REPLICA_ID}
#rm model.test
#gzip ../../param_logs/bestfit_neutral_replica_decode_${REPLICA_ID}
#mv autosomes_deCODE.distr ../../param_logs/bestfit_neutral_replica_decode_${REPLICA_ID}
rm autosomes_deCODE.distr 
#gzip ../../param_logs/bestfit_neutral_replica_decode_${REPLICA_ID}
#mv recParams ../../param_logs/bestfit_neutral_rec_replica_${REPLICA_ID}
#gzip ../../param_logs/bestfit_neutral_rec_replica_${REPLICA_ID}
rm recParams 
rm params 
rm bestfit_neutral_replica_${REPLICA_ID}

cd ..
rmdir ${REPLICA_ID}
#cd ..
#rmdir ${OUTPUTDIR}