gistfile1.txt

#!/bin/sh
#
set -e

check_requirements()
{
  [ ! -f $s2f_bin ] && error "bfast solid2fastq not found." 
  [ ! -f $bwa_s2f ] && error "bwa solid2fastq not found." 
  [ ! -x $s2f_bin ] && error "bfast solid2fastq not executable." 
  [ ! -x $bwa_s2f ] && error "bwa solid2fastq not exectutable." 
  for i in $o_f3 $o_f3q $o_r3 $o_r3q
  do
    [ ! -f $o_f3 ] && error "data file not found: [$i]" 
  done
  return 0
}

log()
{
  echo "$1"
}

error()
{
  echo "$1"
  exit 1
}

diff_this()
{
  f1=$1
  f2=$2
  [ ! -f $f1 ] && error "File not found: $f1" 
  [ ! -f $f2 ] && error "File not found: $f2" 
  if [ ".`diff $f1 $f2`" == "." ]
  then
    out="OK"
  else
    out="ERROR: diff != 0"
  fi

  log "  diff: [$out] $f1 $f2"
}

clean()
{
  rm -rf drd* bf bwa* input*
}

prep()
{
  head -$nl $o_f3 > $f3
  head -$nl $o_r3 > $r3
  head -$nl $o_f3q > $fq3
  head -$nl $o_r3q > $rq3
}

compile()
{
  cd $sdir
  make > /dev/null
  cd - > /dev/null
}

run_bf()
{
  sp=""
  [ ".$2" != "." ] && sp="$2"
  mkdir bf
  cp $1 bf/
  cd bf
  $s2f_bin -b $sp -o prefix $1 
  cd ..
}

run_bwa()
{
  mkdir bwa
  cp $1 bwa/
  cd bwa
  $bwa_s2f "input_" prefix 
  gzip -d *.gz
  cd ..
}

check_split_sizes()
{
  cd bf
  for i in *.fastq
  do
    [ ! -s $i ] && error "  $i: ZERO size!!"
  done
  echo "  size: [OK] (`ls *.fastq | wc -l | sed 's/ //g'`)"
  cd ..
}

# Testing one end
test_one()
{
  log "TEST: No splits. ONE end."
  run_bf  "$f3 $fq3"
  run_bwa "$f3 $fq3"

  diff_this bf/prefix.single.fastq bwa/prefix.single.fastq
}

# Testing two ens
test_two()
{
  log "TEST: No splits. TWO ends."
  run_bf  "$f3 $r3 $fq3 $rq3"
  run_bwa "$f3 $r3 $fq3 $rq3"

  diff_this bf/prefix.single.fastq bwa/prefix.single.fastq
  diff_this bf/prefix.read1.fastq bwa/prefix.read1.fastq
  diff_this bf/prefix.read2.fastq bwa/prefix.read2.fastq
}

# Testing one end (split)
test_one_split()
{
  log "TEST: SPLITS, ONE end."
  run_bf  "$f3 $fq3" "-n$n"
  run_bwa "$f3 $fq3"
  
  cat bf/*.fastq > ./bf.merged.fastq
  diff_this ./bf.merged.fastq bwa/prefix.single.fastq
  rm -f ./bf.merged.fastq

  check_split_sizes
}

test_two_split()
{
  log "TEST: SPLITS, TWO end."
  run_bf  "$f3 $r3 $fq3 $rq3" "-n$n"
  run_bwa "$f3 $r3 $fq3 $rq3"

  cat bf/*read1*fastq     > ./bf.r1.fastq
  cat bf/*read2*fastq     > ./bf.r2.fastq
  cat bf/*single*fastq    > ./bf.single.fastq
  diff_this ./bf.r1.fastq     bwa/prefix.read1.fastq
  diff_this ./bf.r2.fastq     bwa/prefix.read2.fastq
  diff_this ./bf.single.fastq bwa/prefix.single.fastq
  rm -f ./bf.*

  check_split_sizes
}

# main
#
sdir="/Users/drio/projects/bfast_related/versions/dev/bfast/scripts"
s2f_bin="$sdir/solid2fastq"
bwa_s2f="/Users/drio/projects/bwa_related/bio-bwa/trunk/bwa/solid2fastq.pl"

o_f3="./data/bifxTrainingData_10panels_auto/reads/SHIRAZ_20080320_MP_2_Sample1_F3_10panel.csfasta"
o_f3q="./data/bifxTrainingData_10panels_auto/reads/SHIRAZ_20080320_MP_2_Sample1_F3_QV_10panel.qual"
o_r3="./data/bifxTrainingData_10panels_auto/reads/SHIRAZ_20080320_MP_2_Sample1_R3_10panel.csfasta"
o_r3q="./data/bifxTrainingData_10panels_auto/reads/SHIRAZ_20080320_MP_2_Sample1_R3_QV_10panel.qual"

f3="./input_F3.csfasta"
r3="./input_R3.csfasta"
fq3="./input_F3_QV.qual"
rq3="./input_R3_QV.qual"

nl="10001"
n="1000"

check_requirements
compile
clean; prep; test_one
clean; prep; test_two
clean; prep; test_one_split
clean; prep; test_two_split

echo "ALL tests Passed."
clean