0.setup

#!/bin/bash

source globals

clean=1
if [ "$#" -ge 1 ]
then
	if [ "$1" == "clean" ]
	then
		echo "forcing regeneration of all files"
		clean=1
	else
		echo "unknown argument: $1"
		exit 1
	fi
fi

mkdir -p data
cd data

echo downloading gold.fa for chimera detection
if [ ! -e gold.fa ]
then
	wget http://drive5.com/uchime/gold.fa
fi

echo creating symlinks to data
ln -sf ../../2014_*/fasta-qual-mapping-files/*illumina-full.* .
#ln -sf ../../2014_*/fasta-qual-mapping-files/*illumina-mapping2.txt* .

echo making fastq
for run in $RUNS
do
	if [ ! -e $run-full.fastq -o "$clean" -eq 1 ]
	then
		echo making fastq for $run
		echo python $FQ2FASTQ $run-full.fasta $run-full.qual
		python $FQ2FASTQ $run-full.fasta $run-full.qual > $run-full.fastq
	fi
done

FQ=""
echo strip barcode and primer and relabel
for run in $RUNS
do
	if [ ! -e $run-clean.fastq -o "$clean" -eq 1 ]
	then
		awk '{ if (substr($0, 1, 1) != "#") { id = $1; sub("Sample454." , "", id); printf(">%s\n%s\n", id, $2); } }' $run-mapping2.txt > $run-barcodes.fasta
		echo processing $run
		echo python $FQRELABEL $run-full.fastq $PRIMER $run-barcodes.fasta ${run}_
		python $FQRELABEL $run-full.fastq $PRIMER $run-barcodes.fasta ${run}_ > $run-clean.fastq
	fi
	FQ="$FQ $run-clean.fastq"
done

echo combining $FQ into combined.fastq
if [ ! -e combined.fastq -o "$clean" -eq 1 ]
then
	cat $FQ > combined.fastq
fi