comparison imgt_concatenate.sh @ 2:d77d4700fd0a draft

Uploaded

1:b360a373835f

function concat_imgt_files {
	indir=$1
	outdir=$2
	start_line=$3 #line # to start at, 2 to skip header
	id=$4
	cat `find $indir/ -name "1_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/1_Summary.txt"
	cat `find $indir/ -name "2_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/2_IMGT-gapped-nt-sequences.txt"
	cat `find $indir/ -name "3_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/3_Nt-sequences.txt"
	cat `find $indir/ -name "4_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/4_IMGT-gapped-AA-sequences.txt"
	cat `find $indir/ -name "5_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/5_AA-sequences.txt"
	cat `find $indir/ -name "6_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/6_Junction.txt"
	cat `find $indir/ -name "7_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/7_V-REGION-mutation-and-AA-change-table.txt"
	cat `find $indir/ -name "8_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/8_V-REGION-nt-mutation-statistics.txt"
	cat `find $indir/ -name "9_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/9_V-REGION-AA-change-statistics.txt"
	cat `find $indir/ -name "10_*"` | tail -n+${start_line} | awk -v id=$id '{ if($1 !~ /^header/) {$2=$2id; } print}' >> "$outdir/10_V-REGION-mutation-hotspots.txt"
}

echo "Unpacking IMGT file 1.."
imgt_unpack ${inputs[0]} "$workdir/input1"

	echo "Concatenating IMGT file $1..."
	concat_imgt_files "${current_dir}" "$workdir/output" 2 $id
	i=$((i+2))
done


echo "Creating new IMGT zip"
cd "$workdir/output"
tar cfJ "$output" *

2:d77d4700fd0a

function concat_imgt_files {
	indir=$1
	outdir=$2
	start_line=$3 #line # to start at, 2 to skip header
	id=$4
	if [[ "${start_line}" == "1" ]] ; then
		cat `find $indir/ -name "1_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/1_Summary.txt"
		cat `find $indir/ -name "2_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/2_IMGT-gapped-nt-sequences.txt"
		cat `find $indir/ -name "3_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/3_Nt-sequences.txt"
		cat `find $indir/ -name "4_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/4_IMGT-gapped-AA-sequences.txt"
		cat `find $indir/ -name "5_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/5_AA-sequences.txt"
		cat `find $indir/ -name "6_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/6_Junction.txt"
		cat `find $indir/ -name "7_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/7_V-REGION-mutation-and-AA-change-table.txt"
		cat `find $indir/ -name "8_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/8_V-REGION-nt-mutation-statistics.txt"
		cat `find $indir/ -name "9_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/9_V-REGION-AA-change-statistics.txt"
		cat `find $indir/ -name "10_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>1) {$2=$2id; } print}' >> "$outdir/10_V-REGION-mutation-hotspots.txt"
	else 
		cat `find $indir/ -name "1_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/1_Summary.txt"
		cat `find $indir/ -name "2_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/2_IMGT-gapped-nt-sequences.txt"
		cat `find $indir/ -name "3_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/3_Nt-sequences.txt"
		cat `find $indir/ -name "4_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/4_IMGT-gapped-AA-sequences.txt"
		cat `find $indir/ -name "5_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/5_AA-sequences.txt"
		cat `find $indir/ -name "6_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/6_Junction.txt"
		cat `find $indir/ -name "7_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/7_V-REGION-mutation-and-AA-change-table.txt"
		cat `find $indir/ -name "8_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/8_V-REGION-nt-mutation-statistics.txt"
		cat `find $indir/ -name "9_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/9_V-REGION-AA-change-statistics.txt"
		cat `find $indir/ -name "10_*"` | tail -n+${start_line} | awk -F $'\t' -v id=$id 'BEGIN {OFS = FS} { if(NR>0) {$2=$2id; } print}' >> "$outdir/10_V-REGION-mutation-hotspots.txt"
	fi
	
}

echo "Unpacking IMGT file 1.."
imgt_unpack ${inputs[0]} "$workdir/input1"

	echo "Concatenating IMGT file $1..."
	concat_imgt_files "${current_dir}" "$workdir/output" 2 $id
	i=$((i+2))
done

echo "`head $workdir/output/1_Summary.txt`"


echo "Creating new IMGT zip"
cd "$workdir/output"
tar cfJ "$output" *