manyfish

#!/bin/bash

#### Register barracuda executable path
barracudaExec=barracuda
#[ ! -f $barracudaExec ] && echo "Cannot find barracuda executable" && exit 1

#### Clean script
clean()
{
  echo Cleaning . . .
  [ `ps | grep -c execRun.sh` -ne 0 ] && killall execRun.sh
  [ `ps | grep -c barracuda` -ne 0 ] && killall barracuda
  [ `ls | grep -c ^outSai_splited_` -ne 0 ] && rm -f outSai_splited_*
  [ `ls | grep -c ^outSam_splited_` -ne 0 ] && rm -f outSam_splited_*
  [ `ls | grep -c ^splited_` -ne 0 ] && rm -f splited_*
  [ -f devices.txt ] && rm -f devices.txt
  [ -f execList.txt ] && rm -f execList.txt
  [ -f finished ] && rm -f finished
  [ -f execRun.sh ] && rm -f execRun.sh
  [ -f execRun_first.sh ] && rm -f execRun_first.sh
}

#### Trap interrupt
intDetect()
{
  echo Error or break!
  clean
  exit 1
}

#### Trap signal
trap 'intDetect' SIGINT SIGTERM

#### sanity check
[ "$1" == "" -o "$2" == "" -o "$3" == "" ] && echo $0 reference input output [options] && exit 1
options=`echo $* | cut -d" " -f 4-`

#### Gather device information
$barracudaExec deviceQuery > devices.txt

#### gather device information
deviceCount=`wc -l devices.txt | awk '{print $1}'`

#### find and count usable device count
# usable device count
execCount=0;
i=1
[ -f execList.txt ] && rm execList.txt
while [ $i -le $deviceCount ]
do
  curDev=`head -$i devices.txt | tail -1`
  devNum=`echo $curDev| awk '{print $1}'`
  devRam=`echo $curDev| awk '{print $2}'`
  devVer=`echo $curDev| awk '{print $3}'`
  if [ $devRam -ge 512 -a $devVer -ge 12 ] 
  then
    echo Device Number $devNum with RAM=$devRam MB Version=$devVer will be used
    #usable device numbers are recorded in execList.txt
    echo $devNum >> execList.txt
    execCount=$[$execCount+1]
  fi
  i=$[$i+1]
done

#### splitting into files according to usable device count
lineTotal=`wc -l $2 | awk '{print $1}'`

lineDiv4=$[$lineTotal/4]
remainder=$[$lineDiv4%execCount]
if [ $remainder != 0 ]
then 
  lineExec=$[$lineDiv4/$execCount]
  blockSize=$[$[$lineExec+1]*4]
else
  blockSize=$[$lineTotal/$execCount]
fi

echo Total number of lines = $lineTotal, usable devices = $execCount
echo Now spliting \"$2\" with block size = $blockSize
[ `ls | grep -c ^splited_` -ne 0 ] && rm -f splited_* 
split -l $blockSize $2 splited_
splited=`ls splited_*`

### create execRun_first.sh
printf "#!/bin/bash\n" > execRun_first.sh
printf "options=`echo \$* | cut -d\" \" -f 6-`\n" >> execRun_first.sh
printf "echo Using device \$1, \$2 as input, \$3 as reference.\n" >> execRun_first.sh
printf "###Sam output name\n" >> execRun_first.sh
printf "outName='outSam_'\$2\n" >> execRun_first.sh
printf "###SAI output name\n" >> execRun_first.sh
printf "outNameSai='outSai_'\$2\n" >> execRun_first.sh
printf "### Do alignment\n" >> execRun_first.sh
printf "$barracudaExec aln -C \$1 $options \$3 \$2 > \$outNameSai\n" >> execRun_first.sh
printf "[ \$? -ne 0 ] && kill -2 \$4 && exit 1\n" >> execRun_first.sh
printf "echo 1 >> finished\n" >> execRun_first.sh
chmod 755 execRun_first.sh

### create execRun.sh
printf "#!/bin/bash\n" > execRun.sh
printf "options=`echo \$* | cut -d\" \" -f 6-`\n" >> execRun.sh
printf "echo Using device \$1, \$2 as input, \$3 as reference.\n" >> execRun.sh
printf "###Sam output name\n" >> execRun.sh
printf "outName='outSam_'\$2\n" >> execRun.sh
printf "###SAI output name\n" >> execRun.sh
printf "outNameSai='outSai_'\$2\n" >> execRun.sh
printf "### Do alignment\n" >> execRun.sh
printf "$barracudaExec aln -z -C \$1 $options \$3 \$2 > \$outNameSai\n" >> execRun.sh
printf "[ \$? -ne 0 ] && kill -2 \$4 && exit 1\n" >> execRun.sh
printf "echo 1 >> finished\n" >> execRun.sh
chmod 755 execRun.sh

### Submitting jobs to usable devices
[ -f finished ] && rm -f finished 
touch finished
i=1
  curDev=`head -$i execList.txt | tail -1`
  inputName=`echo $splited | awk '{print $1}'`
  ./execRun_first.sh $curDev $inputName $1 $$ $options &
  splited=`echo $splited | cut -d" " -f 2-`
  i=$[$i+1]

while [ $i -le $execCount ]
do
  curDev=`head -$i execList.txt | tail -1`
  inputName=`echo $splited | awk '{print $1}'`
  ./execRun.sh $curDev $inputName $1 $$ $options &
  splited=`echo $splited | cut -d" " -f 2-`
  i=$[$i+1]
done

### Waiting for results
echo Waiting. . .
finishCount=0
while [ $finishCount -lt $execCount ]
do
  sleep 1
  finishCount=`wc -l finished | awk '{print $1}'`
done

### Combine results
outTemp=`ls outSai_splited_*`
echo Combining result $outTemp to \"$3\"
cat outSai_splited_* > $3
clean
echo finished