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

Patch RPS12 detection

Browse Source
This commit is contained in:
Eric Coissac
2025-05-25 10:31:30 +02:00
parent c5b92799b1
commit 9205fd1ed1
4 changed files with 16 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
detectors/cds/bin/do_rps12.sh
View File

@ -31,6 +31,10 @@ else
QUERY="$1" QUERY="$1"
fi fi
shift
GENOME_LENGHT="$2"
if (( $# > 1 )) ; then if (( $# > 1 )) ; then
TEMP=$2 TEMP=$2
else else
@ -45,7 +49,6 @@ AnnotFile="$CDS_DATA_DIR/sp_chlorodb/Annot.lst"
ModelsDir="$CDS_DATA_DIR/sp_chlorodb/models" ModelsDir="$CDS_DATA_DIR/sp_chlorodb/models"
SEQLEN=$(seqlength "${QUERY}") SEQLEN=$(seqlength "${QUERY}")
SEQUENCE=$(readfirstfastaseq "${QUERY}")
pushTmpDir ORG.RPS12 pushTmpDir ORG.RPS12
@ -78,7 +81,10 @@ blastx \
{PREV_CDS = $2;} {PREV_CDS = $2;}
(BEST_EVAL > ($11 + 0.0)) {BEST_EVAL = ($11 + 0.0)} (BEST_EVAL > ($11 + 0.0)) {BEST_EVAL = ($11 + 0.0)}
' > "rps12_locate.hsps" ' \
| $AwkCmd -v glength=${GENOME_LENGHT} \
'!($7 + 0 > glength + 0 && $8 + 0 > glength + 0)' \
> "rps12_locate.hsps"
# #

7
detectors/cds/bin/go_cds.sh
View File

@ -9,6 +9,7 @@
# go_cds.sh <FASTAFILE> [DBROOT] # go_cds.sh <FASTAFILE> [DBROOT]
# #
# - <FASTAFILE> : The fasta file containing the genome to annotate # - <FASTAFILE> : The fasta file containing the genome to annotate
# - <GENOME_LENGHT> : The length of the genome to annotate
# - [DBROOT] : optionnal argument allowing to specify database directory # - [DBROOT] : optionnal argument allowing to specify database directory
# #
# Results are printed to the standard output # Results are printed to the standard output
@ -28,12 +29,14 @@ else
Threads=$ANNOT_MAXCPU Threads=$ANNOT_MAXCPU
fi fi
needarg 1 needarg 2
Fasta=$1; shift Fasta=$1; shift
needfile "$Fasta" needfile "$Fasta"
GenomeLength=$1; shift
# Genome names is set from the base # Genome names is set from the base
# name of the genome file without its extension # name of the genome file without its extension
Genome=$(basename ${Fasta%.*}) Genome=$(basename ${Fasta%.*})
@ -92,7 +95,7 @@ fi
if [[ "$cdsdetection_pass2" == "yes" ]] ; then if [[ "$cdsdetection_pass2" == "yes" ]] ; then
loginfo "running pass2:rps12 exonerate of $Genome on $DbRoot" loginfo "running pass2:rps12 exonerate of $Genome on $DbRoot"
$PROG_DIR/do_rps12.sh $Fasta $temp $PROG_DIR/do_rps12.sh $Fasta $GenomeLength $temp
fi fi
# #

1
detectors/cds/lib/rps12_filter_4.awk
View File

@ -34,6 +34,7 @@ function convert(p1,p2) {
r = $0 r = $0
while (length(s) > 0) { while (length(s) > 0) {
match(s,/[0-9]+\.\.[0-9]+/) match(s,/[0-9]+\.\.[0-9]+/)
if (RLENGTH+0 < 0) break
range = substr(s,RSTART,RLENGTH) range = substr(s,RSTART,RLENGTH)
s = substr(s,RSTART+RLENGTH+1) s = substr(s,RSTART+RLENGTH+1)
match(range,/^[0-9]+/) match(range,/^[0-9]+/)

3
org-annotate.sh
View File

@ -441,6 +441,7 @@ do
shift shift
done done
loginfo "Locus tag prefix provided: $tagprefix" loginfo "Locus tag prefix provided: $tagprefix"
loginfo "Locus tag numbered from..: $locusshift" loginfo "Locus tag numbered from..: $locusshift"
loginfo "NCBI taxid provided......: $taxid" loginfo "NCBI taxid provided......: $taxid"
@ -572,7 +573,7 @@ pushTmpDir ORG.organnot
cdsdetection_pass1=$cdsdetection_pass1 \ cdsdetection_pass1=$cdsdetection_pass1 \
cdsdetection_pass2=$cdsdetection_pass2 \ cdsdetection_pass2=$cdsdetection_pass2 \
cdsdetection_pass3=$cdsdetection_pass3 \ cdsdetection_pass3=$cdsdetection_pass3 \
${PROG_DIR}/detectors/cds/bin/go_cds.sh "${RESULTS}.norm.fasta" >> "${RESULTS}.annot" ${PROG_DIR}/detectors/cds/bin/go_cds.sh "${RESULTS}.norm.fasta" "$sl" >> "${RESULTS}.annot"
loginfo "Done." loginfo "Done."
fi fi