Mobster

#!/bin/bash

# job name
#SBATCH --job-name=mobster

# How much memory do you need **per core**?
#SBATCH --mem-per-cpu=10G

# time to run
#SBATCH --time=0-6:00

# Number of cores you need
#SBATCH --ntasks=1

# output name
#SBATCH --output=~/mobster_results/mobster.out

export MOBPATH=~/mobster_results/resources/mobiome/

# psecify the input bam file
input=LP6005878-DNA_C07

# run picardtools to find insert size metrics
picard CollectInsertSizeMetrics \
I=~/mobster_results/$input.bam \
O=~/mobster_results/$input_sizemets.txt \
H=~/mobster_results/$input_inshist.pdf \
M=0.2

# extract the insert summary stat lengths for use with mobster properties file. 
mean=$(cat $input_sizemets.txt | head -8 | tail -1 | awk '{print $6}'| awk -F '.' '{print $1}')
sd=$(cat $input_sizemets.txt | head -8 | tail -1 | awk '{print $7}'| awk -F '.' '{print $1}')
length_99=$(cat $input_sizemets.txt | head -8 | tail -1 | awk '{print $20}'| awk -F '.' '{print $1}')

echo $mean
echo $sd
echo $length_99

# input the summar stats into the properties file to run with mobster
sed "s/MEAN_FRAGMENT_LENGTH=/MEAN_FRAGMENT_LENGTH=$mean/g" mobster_props.properties > mobster_props_$input_int1.properties
sed "s/SD_FRAGMENT_LENGTH=/SD_FRAGMENT_LENGTH=$sd/g" mobster_props_$input_int1.properties > mobster_props_$input_int2.properties
sed "s/LENGTH_99PROCENT_OF_FRAGMENTS=/LENGTH_99PROCENT_OF_FRAGMENTS=$length_99/g" mobster_props_$input_int2.properties > mobster_props_$input.properties
rm mobster_props_$input_int1.properties
rm mobster_props_$input_int2.properties

#run mobster using the modified properties file (which is publicly available with mobster) and the input bam. 
mobster -properties mobster_props_$input.properties -in $input.bam -out ~/mobster_results/$input_newfa -sn $input_newfa_sn