makeblastdb.sh

#!/bin/bash

# Script for Biomaj PostProcess
# author : ofilangi, osallou
# date   : 19/06/2007
# update : 22/10/2010 fix bug in generated alias file + a few cleanups
#          23/12/2015 use makeblastdb for ncbi blast+
#
#  -title  Title for database file [String]  Optional
#  -in  Input file(s) for formatting [File In]  Optional
#  -logfile  Logfile name: [File Out]  Optional
#    default = makeblastdb.log
#  -dbtype nucl
#  -parse_seqids
#
source activate blast
#----------
#GLOBAL DEF
#----------
BLASTDB_DIR="$datadir/index-blast"; # Path where aliases files should be generated
mkdir -p $BLASTDB_DIR
MAKEBLASTDB="makeblastdb"; # Path to makeblastdb executable


#----------
# FUNCTIONS 
#----------
# createAlias: builds an alias file
# arg1: file to write to
# arg2: bank name
# arg3: db file list
createAlias() {
  local file=$1;
  local nomBanque=$2;
  local lFiles=$3;

  rm -f $file;
  echo "#" > $file
  echo "# Alias file created "`date` >>$file 
  echo "#" >>$file ;
  echo "#">> $file ;
  echo "TITLE "$nomBanque >> $file; 
  echo "#" >> $file;
  echo "DBLIST "$lFiles >>$file;
  echo "#" >> $file;
  echo "#GILIST" >> $file;
  echo "#" >> $file;
  echo "#OIDLIST" >> $file;
  echo "#" >> $file;
}

#-----
# MAIN
#-----

if (test $# -ne 4) then
  echo "arguments:" 1>&2
  echo "1: input files"
  echo "2: working directory" 1>&2
  echo "3: makeblastdb options (without -in for input file)" 1>&2
  echo "4: bank name" 1>&2
  echo `makeblastdb -help`;
  exit -1
fi

relWorkDir=`echo "$2" | sed "s/\/*$//"` # remove useless trailing slash

workdir=$datadir/$dirversion/future_release
workdir=$workdir/$relWorkDir;

rm -rf $workdir;
mkdir -p $workdir ;

if ( test $? -ne 0 ) then
  echo "Cannot create $workdir." 1>&2 ;
  exit 1;
fi

cd $workdir

# Some vars for links creation
back="";
dir=$relWorkDir;
OLDIFS=$IFS;
IFS="/";
for i in $dir
do
  back="../"$back;
done
IFS=$OLDIFS;

# Create links to input files into the working dir
listFile="";

for expression in $1
do
  # the basename can be a regex
  lsFile=`ls $datadir/$dirversion/future_release/$expression`;
  if ( test $? -ne 0 ) then
    echo "No input file found in dir `pwd`." 1>&2 ;
    exit 1
  fi
  baseFile=`dirname $expression`;
  for f in $lsFile
  do
    name=`basename $f`;
    rm -f $4.p*;
    rm -f $4.n*;
    nameLink=`echo $name | cut -d"." -f1`;
    ln -s $back/$baseFile/$name $nameLink;
    if ( test $? -ne 0 ) then
      echo "Cannot create link. [ln -s $back$f $name]" 1>&2 ;
      exit 1
    fi
    if (test -z "$listFile") then
      listFile=$nameLink;
    else
      listFile=$nameLink" "$listFile;
    fi
  done
done

echo "Input sequence file list: $listFile";

if (test -z "$listFile") then
  echo "No input file found." 1>&2 ;
  exit 1
fi

nameB=$4;
echo "Database name: $nameB";

echo "Working in "`pwd`;
echo "Launching makeblastdb [makeblastdb -in $listFile $3 -out $nameB]";

# Execute makeblastdb
$MAKEBLASTDB -in "$listFile" $3 -out $nameB;

makeblastdbResult=$?
if ( test $makeblastdbResult -ne 0 ) then
  echo "Makeblastdb failed with status $makeblastdbResult" 1>&2 ;
  exit 1
fi

echo "##BIOMAJ#blast###$2$nameB"

# Delete temp files and links
#-------------------------------------------------------------
ls -ltrh
rm -f $listFile;
rm -f makeblastdb.log

# Add generated files to biomaj postprocess dependance
echo "Generated files:";
for ff in `ls *`
do
  echo $PP_DEPENDENCE$PWD/$ff;
done

goodPath=`readlink $datadir/$dirversion/future_release -s -n`;
if ( test $? -ne 0 ) then
  echo "Failed to get version path: readlink returned with an error [$goodPath]" 1>&2 ;
  exit 1
fi

# Search for nal files which are sometimes generated by makeblastdb.
lsAl=`ls *.?al 2> /dev/null`;

if ( test $? -ne 0 ) then
  echo "No alias file found.";
  lsAl="";
else
  echo "Generated alias files:"
  echo "$lsAl";
fi

# If nal files were generated, use them to generate nal files in $BLASTDB_DIR
for fileIndexVirtuel in $lsAl 
do
  echo "Found alias file: [$fileIndexVirtuel]";
  listIndex=`cat $fileIndexVirtuel | grep DBLIST`;
  listFile2="";
  for f in $listIndex
  do
    if (test $f != "DBLIST") then
      listFile2=$goodPath/$relWorkDir/$f" "$listFile2;
    fi
  done
  echo "Creating alias in [$BLASTDB_DIR/$fileIndexVirtuel]";
  createAlias $BLASTDB_DIR/$fileIndexVirtuel $nameB "$listFile2"
done

# Else, if no nal file was generated by makeblastdb, create them
if (test -z "$lsAl") then
  ext=`ls | grep .*hr$ | tail -c5 | head -c2`al;
  echo "Creating alias file [$PWD/$4$ext]";
  
  listNhr=`ls *.*hr | sed 's/\..hr$//g'`;
  listFileNalRel=""; # List of blast db files, relative path
  listFileNalAbs=""; # List of blast db files, absolute path
  for f in $listNhr
  do
    listFileNalRel=$f" "$listFileNalRel;
    listFileNalAbs=$goodPath/$relWorkDir/$f" "$listFileNalAbs;
  done

  createAlias $4$ext $nameB "$listFileNalRel";
  echo $PP_DEPENDENCE$PWD/$4$ext;
  
  echo "Creating alias in [$BLASTDB_DIR/$4$ext]";
  createAlias $BLASTDB_DIR/$4$ext $nameB "$listFileNalAbs" ;
fi

deactivate