run-hmmer.sh

run-hmmer.sh

#!/bin/sh
# run hmmer using hmm files
#INDIR=$1 #directory to search for .fastq.gz files

INDIR=~/work/NPRB

cd ~/work/hmm

for hmmfile in *.hmm; do
	hmm=`basename $hmmfile .hmm`;
	echo $hmm
	ls -l $INDIR
	for seqfile in $INDIR/*fastq.gz; do
		basefile=`basename $seqfile .fastq.gz`;
		basedir=`dirname $seqfile`
		echo $basefile;
#		~/apps/bbmap/reformat.sh in=$seqfile out1=$basedir/$basefile.R1.fasta.gz out2=$basedir/$basefile.R2.fasta.gz fastawrap=0
		if [ ! -e $basedir/$basefile.R1.faa.gz ]; then
		~/apps/orfm/orfm $basedir/$basefile.R1.fasta.gz | gzip -c > $basedir/$basefile.R1.faa.gz
		~/apps/orfm/orfm $basedir/$basefile.R2.fasta.gz | gzip -c > $basedir/$basefile.R2.faa.gz
		fi;
		~/apps/hmmer/binaries/hmmsearch -o txt/$hmm.$basefile.R1.txt --pfamtblout pfam/$hmm.$basefile.R1.pfam --max -A aln/$hmm.$basefile.R1.aln --tblout tsv/$hmm.$basefile.R1.tsv --notextw --cpu 3  -Z 1000000 -E 0.1 $hmmfile $basedir/$basefile.R1.faa.gz &
		~/apps/hmmer/binaries/hmmsearch -o txt/$hmm.$basefile.R2.txt --pfamtblout pfam/$hmm.$basefile.R2.pfam --max -A aln/$hmm.$basefile.R2.aln --tblout tsv/$hmm.$basefile.R2.tsv --notextw --cpu 4  -Z 1000000 -E 0.1 $hmmfile $basedir/$basefile.R2.faa.gz &
		wait
	done;
done;