add a -l option to set a minimum length on the contig to annotate

org-annotate.sh

@@ -25,6 +25,7 @@ irdetection="yes"
 organism="no"
 types="chloro"
 partial=0
+minlength=0
 
 function usage {
 	echo "Usage:" ;  
@@ -56,6 +57,9 @@ function usage {
 	echo
 	echo '  -p     | --partial'
 	echo '      Indicates that the genome sequence is partial and therefore in several contigs'
+	echo
+	echo '  -l     | --min-length'
+	echo '      Indicates for partial mode the minimum length of contig to annotate'
    	exit $2
 }
 
@@ -67,7 +71,7 @@ function fastaIterator() {
 }
 
 # options may be followed by one colon to indicate they have a required argument
-if ! options=$(getopt -o t:o:icrmhp -l ncbi-taxid:,organism,no-ir-detection,chloroplast,nuclear-rdna,mitochondrion,partial,help -- "$@")
+if ! options=$(getopt -o t:o:icrmhpl: -l ncbi-taxid:,organism,no-ir-detection,chloroplast,nuclear-rdna,mitochondrion,partial,min-length:,help -- "$@")
 then
     # something went wrong, getopt will put out an error message for us
     usage $0 1
@@ -85,6 +89,7 @@ do
     -r|--nuclear-rdna) types="nucrdna" ;;
     -m|--mitochondrion) types="mito" ;;
     -p|--partial) partial="1" ;;
+    -l|--min-length) minlength="$2" ; shift ;;
     -h|--help)  usage $0 0;;
     (--) shift; break;;
     (-*) echo "$0: error - unrecognized option $1" 1>&2; exit 1;;
@@ -97,6 +102,7 @@ loginfo "Annotating mode.....: $types"
 loginfo "IR detection mode...: $irdetection"
 loginfo "Organism............: $organism"
 loginfo "Partial mode........: $partial"
+loginfo "Minimum length......: $minlength"
 
 
 #############################
@@ -124,6 +130,9 @@ pushTmpDir ORG.organnot
 		unset IFS
 		if [[ ! -z "${sequence}" ]] ; then
 			echo "${sequence}" > toannotate.fasta
+			sl=$(seqlength "toannotate.fasta")
+			
+			if (( sl >= minlength )) ; then
 			
 				seqid=$($AwkCmd '(NR==1) {print substr($1,2,1000)}' toannotate.fasta)
 				
@@ -326,7 +335,8 @@ pushTmpDir ORG.organnot
 				loginfo "Closing sequence part..."
 					echo "//"
 				loginfo "Done."
-		fi
+			fi # End of the minimum length condition
+		fi  # End of not empty sequence condition
 		
 		IFS=$'\f'
 	done # End of the loop over the sequences