PhyloSkims/annotate
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Allows for specifying a start point for the locus tag numbering

Browse Source 
Former-commit-id: cf32f0ed629e22a8d3434e29b55f10b0ac275a78
Former-commit-id: d86feaf4dfe9622b7f761bb5facb249aa61ae3b4
This commit is contained in:
Eric Coissac
2022-02-14 15:10:03 +01:00
parent d56aeaf698
commit 05551549bf
1 changed files with 22 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
org-annotate.sh
View File

@ -43,6 +43,8 @@ cdsdetection_pass2="yes"
trnadetection="yes" trnadetection="yes"
rrnadetection="yes" rrnadetection="yes"
idprefix="no" idprefix="no"
tagprefix="no"
locusshift=1
organism="no" organism="no"
country="no" country="no"
specimen="no" specimen="no"
@ -102,6 +104,15 @@ function usage {
echo ' to build the complete id. This id is used only if' echo ' to build the complete id. This id is used only if'
echo ' an ENA project is specified.' echo ' an ENA project is specified.'
echo echo
echo ' -L | --locus-prefix <prefix>'
echo ' Prefix used to build the locus tag of every annotated genes'
echo ' generated locus tags follow the pattern : prefix_###,'
echo ' where ### is a number following the order of gene in the embl file'
echo ' starting at locus tag shift (default 1).'
echo
echo ' -S | --locus-shift <###>'
echo ' Start number for building locus tags'
echo
echo ' Annotation of partial sequences' echo ' Annotation of partial sequences'
echo ' -p | --partial' echo ' -p | --partial'
echo ' Indicates that the genome sequence is partial and therefore in several contigs' echo ' Indicates that the genome sequence is partial and therefore in several contigs'
@ -194,7 +205,7 @@ function fastaIterator() {
} }
# options may be followed by one colon to indicate they have a required argument # options may be followed by one colon to indicate they have a required argument
if ! options=$(getopt -o s:t:o:b:P:i:fcrmhpl:NICDETR -l specimen:,ncbi-taxid:,organism:,country:,project:,id-prefix:,not-force-ncbi,chloroplast,nuclear-rdna,mitochondrion,partial,min-length:,help,no-normalization,no-ir-detection,no-cds,no-cds-pass1,no-cds-pass2,no-trna,no-rrna -- "$@") if ! options=$(getopt -o s:t:o:b:P:i:fcrmhpl:NICDETRL:S: -l specimen:,ncbi-taxid:,organism:,country:,project:,id-prefix:,not-force-ncbi,chloroplast,nuclear-rdna,mitochondrion,partial,min-length:,help,no-normalization,no-ir-detection,no-cds,no-cds-pass1,no-cds-pass2,no-trna,no-rrna,locus-prefix:,locus-shift: -- "$@")
then then
# something went wrong, getopt will put out an error message for us # something went wrong, getopt will put out an error message for us
usage $0 1 usage $0 1
@ -214,6 +225,8 @@ do
-o|--organism) organism="$2" ; shift ;; -o|--organism) organism="$2" ; shift ;;
-P|--project) project="$2" ; shift ;; -P|--project) project="$2" ; shift ;;
-i|--id-prefix) idprefix="$2" ; shift ;; -i|--id-prefix) idprefix="$2" ; shift ;;
-L|--locus-prefix) tagprefix="$2" ; shift ;;
-S|--locus-shift) locusshift="$2" ; shift ;;
-f|--not-force-ncbi) resetorganism="no" ;; -f|--not-force-ncbi) resetorganism="no" ;;
-p|--partial) partial="1" ;; -p|--partial) partial="1" ;;
-l|--min-length) minlength="$2" ; shift ;; -l|--min-length) minlength="$2" ; shift ;;
@ -232,6 +245,8 @@ do
shift shift
done done
loginfo "Locus tag prefix provided: $tagprefix"
loginfo "Locus tag numbered from..: $locusshift"
loginfo "NCBI taxid provided......: $taxid" loginfo "NCBI taxid provided......: $taxid"
if [[ "$taxid" != "no" ]] ; then if [[ "$taxid" != "no" ]] ; then
@ -500,14 +515,14 @@ pushTmpDir ORG.organnot
print line}' > "${RESULTS}.sorted.annot" print line}' > "${RESULTS}.sorted.annot"
loginfo "Done." loginfo "Done."
if [[ "$idprefix" != "no" ]] ; then if [[ "$tagprefix" != "no" ]] ; then
loginfo "Adding locus tags..." loginfo "Adding locus tags from number: $locusshift..."
cat "${RESULTS}.sorted.annot" \ cat "${RESULTS}.sorted.annot" \
| $AwkCmd -v idprefix="$idprefix" ' | $AwkCmd -v tagprefix="$tagprefix" \
BEGIN {n=1} -v locusshift="$locusshift" '
/^FT +\/locus_tag=""/ { /^FT +\/locus_tag=""/ {
sub(/locus_tag=""/,"locus_tag=\""idprefix"_"n"\"",$0); sub(/locus_tag=""/,"locus_tag=\""tagprefix"_"locusshift"\"",$0);
n++; locusshift++;
} }
{ {
print $0 print $0