Bug fixes

- Added check if db directory is present.
- db files are automatically extracted if db directory is missing.
- Blast SCG search does not require write permissions in the
installation directory any more.

DAS_Tool

@@ -34,15 +34,15 @@ thisdir=$(pwd)
 
 function display_version() {
  echo
- echo "DAS Tool version 1.1"
+ echo "DAS Tool version 1.1.0"
  echo
  exit 1
 }
 
 
 function display_help() {
  echo " "
- echo "DAS Tool version 1.1"
+ echo "DAS Tool version 1.1.0"
  echo " "
  echo "Usage: DAS_Tool -i methodA.scaffolds2bin,...,methodN.scaffolds2bin"
  echo " -l methodA,...,methodN -c contigs.fa -o myOutput"
@@ -142,8 +142,7 @@ while [ "$1" != "" ]; do
  --megabin_penalty ) shift
  c_weight=$1
  ;;
- --debug ) shift
- debug="TRUE"
+ --debug ) debug="TRUE"
  ;;
  -h | --help ) display_help
  exit
@@ -206,14 +205,14 @@ fi
 if [ ! -f "$contigs" ]
 then
  echo contig file not found: $contigs
- echo stopping
+ echo Aborting
  exit 1
 fi
 
 if [ ! -z "$proteins" ] && [ ! -f "$proteins" ]
 then
  echo proteins file not found: $proteins
- echo stopping
+ echo Aborting
  exit 1
 fi
 
@@ -222,13 +221,25 @@ do
 if [ ! -f $i ];
 then
  echo "scaffolds2bin file not found: $i "
+ echo Aborting
  exit 1
 fi
 done
 
 # Existence and permissions of output directory
-if [ ! -d $(dirname $outputbasename) ]; then echo "Directory for output files does not exist: $(dirname $outputbasename). Aborting"; exit 1; fi
-if [ ! -w $(dirname $outputbasename) ] ; then echo "Can't write in output folder: $(dirname $outputbasename). Permission denied. Aborting"; exit 1; fi
+if [ ! -d $(dirname "$outputbasename") ]; then echo "Directory for output files does not exist: $(dirname $outputbasename). Aborting"; exit 1; fi
+if [ ! -w $(dirname "$outputbasename") ] ; then echo "Can't write in output folder: $(dirname $outputbasename). Permission denied. Aborting"; exit 1; fi
+
+# Existence of database directory
+if [ ! -d $DIR\/db ]; then 
+echo "Database directory does not exist: $(dirname $DIR\/db)."
+echo "Trying to extract $DIR\/db.zip";
+unzip -o $DIR\/db.zip -d $DIR
+if [ ! -d $DIR\/db ]; then 
+echo "Extraction failed. Aborting."
+exit 1; 
+fi
+fi
 
 
 #
@@ -244,7 +255,7 @@ then
  proteins=$outputbasename\_proteins.faa
 
  echo "predicting genes using prodigal"
-# grep ">" $contigs | perl -pe 's/>//g;' | shuf > $outputbasename\_dastoolthreadstmpheader
+ #grep ">" $contigs | perl -pe 's/>//g;' | shuf > $outputbasename\_dastoolthreadstmpheader
  grep ">" $contigs | sed 's/>//g;' | shuf > $outputbasename\_dastoolthreadstmpheader
  dastoolthreadstmpheader=$outputbasename\_dastoolthreadstmpheader
  total_lines=$(wc -l <${dastoolthreadstmpheader})
@@ -264,7 +275,7 @@ then
 elif [ ! -f $proteins ]
 then
  echo protein file not found: $proteins
- echo stopping
+ echo Aborting
  exit 1
 fi
 
@@ -292,6 +303,11 @@ else
  echo skipping single copy gene identification
 fi
 
+if [ ! -f $bscg ] || [ ! -f $ascg ]; then
+echo "single copy gene prediction using $search_engine failed. Aborting"
+ exit 1
+fi
+
 
 
 #

src/scg_blank_blast.rb

@@ -16,8 +16,9 @@
 input_file = ARGV[1]
 output_dir = File.dirname(input_file)
 
-datab = ARGV[2]
-puts "database name of all proteins is #{datab}"
+datab_in = ARGV[2]
+datab_out = "#{File.dirname ARGV[1]}" + "/" + "#{File.basename ARGV[2]}"
+puts "database name of all proteins is #{datab_in}"
 
 db_name = ARGV[3]
 puts "database name of SCGs is #{db_name}"
@@ -28,8 +29,8 @@
 threads = ARGV[5]
 
 #build databases
-full_db = system "makeblastdb -in #{datab} -dbtype prot"
-abort "makeblastdb did not work for #{datab}, please check your input file" unless full_db
+full_db = system "makeblastdb -in #{datab_in} -dbtype prot -out #{datab_out}"
+abort "makeblastdb did not work for #{datab_in}, please check your input file" unless full_db
 
 # find SCG candidates
 puts "finding SCG candidates..."
@@ -49,9 +50,9 @@
 # verify SCGs by blasting against all proteins of all genomes
 puts "verifying selected SCGs..."
 db_blast_out = File.join(output_dir,File.basename(input_file) + ".all.b6")
-db_blast_ok = system "blastp -db #{datab} -query #{scg_candidates} -outfmt '6 qseqid sseqid pident length qlen slen evalue bitscore' -evalue 0.00001 -out #{db_blast_out} -max_target_seqs 1 -num_threads #{threads}"
+db_blast_ok = system "blastp -db #{datab_out} -query #{scg_candidates} -outfmt '6 qseqid sseqid pident length qlen slen evalue bitscore' -evalue 0.00001 -out #{db_blast_out} -max_target_seqs 1 -num_threads #{threads}"
 abort "verifying blast did not work" unless db_blast_ok
-system "rm #{datab}.psq #{datab}.pin #{datab}.phr"
+system "rm #{datab_out}.psq #{datab_out}.pin #{datab_out}.phr"
 puts "starting annotations of single copy cogs..."
 
 # Read db_lookup