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

Patch a bug in the final sequence formating occuring when the input

Browse Source 
sequence has not 60 char per line

Former-commit-id: 213735f5b9f3cd817053e284d7844cfdd69726c6
Former-commit-id: 074b4aaac0eac00de9b3b48e75804417ce780a2d
This commit is contained in:
Eric Coissac
2018-01-18 21:58:50 +01:00
parent 04ea0f110d
commit 08d7c940a4
2 changed files with 16 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@ -46,6 +46,7 @@ BEGIN {
Entry[NbEntry]["to"] = $5+0 Entry[NbEntry]["to"] = $5+0
Entry[NbEntry]["score"] = $6+0 Entry[NbEntry]["score"] = $6+0
Entry[NbEntry]["strand"] = $7 Entry[NbEntry]["strand"] = $7
Entry[NbEntry]["stop"] = (Align !~ "\\*")
if (valid) { if (valid) {
for (i = $4+0 ; i <= $5+0; i++) for (i = $4+0 ; i <= $5+0; i++)
Cover[i] = 1 Cover[i] = 1

19
org-annotate.sh
View File

@ -93,6 +93,12 @@ do
shift shift
done done
loginfo "Annotating mode.....: $types"
loginfo "IR detection mode...: $irdetection"
loginfo "Organism............: $organism"
loginfo "Partial mode........: $partial"
############################# #############################
pushTmpDir ORG.organnot pushTmpDir ORG.organnot
@ -125,7 +131,7 @@ pushTmpDir ORG.organnot
chloro) chloro)
loginfo "Annotating a plant chloroplast genome..." loginfo "Annotating a plant chloroplast genome..."
if [[ "$irdetection"=="yes" ]] && (( partial == 0 )) ; then if [[ "$irdetection" == "yes" ]] && (( partial == 0 )) ; then
loginfo "Normalizing the structure of the Chloroplast sequence..." loginfo "Normalizing the structure of the Chloroplast sequence..."
loginfo " LSC + IRB + SSC + IRA" loginfo " LSC + IRB + SSC + IRA"
@ -137,6 +143,7 @@ pushTmpDir ORG.organnot
loginfo "Done." loginfo "Done."
else else
loginfo "No normalization of the structure of the Chloroplast sequence..."
cat toannotate.fasta > "${RESULTS}.norm.fasta" cat toannotate.fasta > "${RESULTS}.norm.fasta"
rm -f "${RESULTS}.annot" rm -f "${RESULTS}.annot"
touch "${RESULTS}.annot" touch "${RESULTS}.annot"
@ -167,7 +174,7 @@ pushTmpDir ORG.organnot
loginfo "Annotating a plant rDNA cistron..." loginfo "Annotating a plant rDNA cistron..."
loginfo "Normalizing the structure of the cistron sequence..." loginfo "Normalizing the structure of the cistron sequence..."
${PROG_DIR}/detectors/normalizerdna/bin/go_normalizerdna.sh toannotate.fasta > "${RESULTS}.norm.fasta" ${PROG_DIR}/detectors/normalizerdna/bin/go_normalizerdna.sh toannotate.fasta > "${RESULTS}.norm.fasta"
loginfo "Done." loginfo "Done."
loginfo "Annotating the rRNA genes..." loginfo "Annotating the rRNA genes..."
@ -232,7 +239,7 @@ pushTmpDir ORG.organnot
echo "FT /organelle=\"mitochondrion\"" echo "FT /organelle=\"mitochondrion\""
;; ;;
*) *)
loginfo "Nuclear sequence" loginfo "Nuclear sequence"
;; ;;
esac esac
@ -297,6 +304,10 @@ pushTmpDir ORG.organnot
loginfo "Reformating sequences..." loginfo "Reformating sequences..."
lines=$(wc -l "${RESULTS}.norm.fasta" | $AwkCmd '{print $1}') lines=$(wc -l "${RESULTS}.norm.fasta" | $AwkCmd '{print $1}')
loginfo "Sequence length $(seqlength ${RESULTS}.norm.fasta)"
loginfo "lines $lines"
formatfasta "${RESULTS}.norm.fasta" | \
$AwkCmd -v lines=$lines ' \ $AwkCmd -v lines=$lines ' \
! /^>/ { \ ! /^>/ { \
seq=tolower($0); \ seq=tolower($0); \
@ -311,7 +322,7 @@ pushTmpDir ORG.organnot
if (NR==lines) \ if (NR==lines) \
{pos-=1}; \ {pos-=1}; \
printf(" %6d\n",pos) \ printf(" %6d\n",pos) \
}' "${RESULTS}.norm.fasta" }'
loginfo "Done." loginfo "Done."
loginfo "Closing sequence part..." loginfo "Closing sequence part..."