doMast

#!/bin/bash

# Given a motif file (meme.txt) and a directory full of genomes in fasta format
# launches mast on each genome

if [ "$#" -lt 6 ]; then
    echo 'doMast MOTIF_FILE FASTA_DIR BFILE_DIR GENBANK_DIR RESULTS_DIR REGULATOR_NAME'
    exit 65
fi

MOTIF_FILE=$1
FASTA_DIR=$2
BFILE_DIR=$3
GENBANK_DIR=$4
RESULTS_DIR=$5
REGULATOR_NAME=$6

for i in $(ls $FASTA_DIR)
do
	org=$(echo $i | awk 'BEGIN { FS = "." } ; {print $1}')
	mast $MOTIF_FILE $FASTA_DIR/$i -oc $RESULTS_DIR/$org'_'$REGULATOR_NAME -bfile $BFILE_DIR/$org.ntfreq -ev 100
	mast $MOTIF_FILE $FASTA_DIR/$i -oc $RESULTS_DIR/$org'_'$REGULATOR_NAME -hit_list -bfile $BFILE_DIR/$org.ntfreq -ev 100 > $RESULTS_DIR/$org'_'$REGULATOR_NAME'/rawhits.txt'
	./getMastEvalue $RESULTS_DIR/$org'_'$REGULATOR_NAME'/rawhits.txt' $FASTA_DIR/$i > $RESULTS_DIR/$org'_'$REGULATOR_NAME'/hits.txt'
	./parseMast -l -t 1e-4 -o $RESULTS_DIR/$org'_'$REGULATOR_NAME/hits.txt $GENBANK_DIR/$org.gbk $REGULATOR_NAME > $RESULTS_DIR/$org'_'$REGULATOR_NAME'_orientation.tab'
	./parseMast -l -t 1e-4 $RESULTS_DIR/$org'_'$REGULATOR_NAME/hits.txt $GENBANK_DIR/$org.gbk $REGULATOR_NAME > $RESULTS_DIR/$org'_'$REGULATOR_NAME'_noorientation.tab'
done