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

Switch from awk to gawk

Browse Source 
Former-commit-id: 70b104ab89d4c90a35925dbbcacdafaf6c1f2836
Former-commit-id: bf29abf66345cbc0ca81461064dcca63dfd4c15c
This commit is contained in:
Eric Coissac
2015-11-08 19:33:00 +01:00
parent 6fa63daf3f
commit 2261fa1c48
16 changed files with 52 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
detectors/normalize/tools/buildSCDB.sh
View File

@ -33,7 +33,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Building LSC coorientation graph..."
${PROG_DIR}/coorienteSC.sh LSC.fasta 20000 ${ORG_LOGFILE} > LSC.tgf
${PROG_DIR}/cc.py LSC.tgf > LSC.cc
loginfo " --> $(awk '{print $1}' LSC.cc | uniq | wc -l) connected componants"
loginfo " --> $($AwkCmd '{print $1}' LSC.cc | uniq | wc -l) connected componants"
loginfo "Done"
@ -46,7 +46,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Extracting main connected components for LCS..."
rm -f LSC.direct.fasta
touch LSC.direct.fasta
for id in `awk '($1==0) {print $2}' LSC.cc`; do
for id in `$AwkCmd '($1==0) {print $2}' LSC.cc`; do
fastafetch -f LSC.fasta -i LSC.index -q "${id}" >> LSC.direct.fasta
done
loginfo " --> $(fastaCount LSC.direct.fasta) sequences"
@ -57,7 +57,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Extracting second connected components for LCS..."
rm -f LSC.reverse.fasta
touch LSC.reverse.fasta
for id in `awk '($1==1) {print $2}' LSC.cc`; do
for id in `$AwkCmd '($1==1) {print $2}' LSC.cc`; do
fastafetch -f LSC.fasta -i LSC.index -q "${id}" >> LSC.reverse.fasta
done
loginfo " --> $(fastaCount LSC.reverse.fasta) sequences"
@ -75,7 +75,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Checking LCS homogeneity..."
${PROG_DIR}/coorienteSC.sh LSC.direct.fasta 20000 ${ORG_LOGFILE} > LSC_RefDB.tgf
${PROG_DIR}/cc.py LSC_RefDB.tgf > LSC_RefDB.cc
NCC=$(awk '{print $1}' LSC_RefDB.cc | uniq | wc -l)
NCC=$($AwkCmd '{print $1}' LSC_RefDB.cc | uniq | wc -l)
if (( $NCC == 1 )); then
loginfo " --> $NCC connected componants"
else
@ -105,7 +105,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Building SSC coorientation graph..."
${PROG_DIR}/coorienteSC.sh SSC.fasta 5000 ${ORG_LOGFILE} > SSC.tgf
${PROG_DIR}/cc.py SSC.tgf > SSC.cc
loginfo " --> $(awk '{print $1}' SSC.cc | uniq | wc -l) connected componants"
loginfo " --> $($AwkCmd '{print $1}' SSC.cc | uniq | wc -l) connected componants"
loginfo "Done"
@ -119,7 +119,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Extracting main connected components for SSC..."
rm -f SSC.direct.fasta
touch SSC.direct.fasta
for id in `awk '($1==0) {print $2}' SSC.cc`; do
for id in `$AwkCmd '($1==0) {print $2}' SSC.cc`; do
fastafetch -f SSC.fasta -i SSC.index -q "${id}" >> SSC.direct.fasta
done
loginfo " --> $(fastaCount SSC.direct.fasta) sequences"
@ -130,7 +130,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Extracting second connected components for SSC..."
rm -f SSC.reverse.fasta
touch SSC.reverse.fasta
for id in `awk '($1==1) {print $2}' SSC.cc`; do
for id in `$AwkCmd '($1==1) {print $2}' SSC.cc`; do
fastafetch -f SSC.fasta -i SSC.index -q "${id}" >> SSC.reverse.fasta
done
loginfo " --> $(fastaCount SSC.reverse.fasta) sequences"
@ -148,7 +148,7 @@ pushTmpDir ORG.buildSCDB
loginfo "Checking SSC homogeneity..."
${PROG_DIR}/coorienteSC.sh SSC.direct.fasta 5000 ${ORG_LOGFILE} > SSC_RefDB.tgf
${PROG_DIR}/cc.py SSC_RefDB.tgf > SSC_RefDB.cc
NCC=$(awk '{print $1}' SSC_RefDB.cc | uniq | wc -l)
NCC=$($AwkCmd '{print $1}' SSC_RefDB.cc | uniq | wc -l)
if (( $NCC == 1 )); then
loginfo " --> $NCC connected componants"
else

6
detectors/normalize/tools/coorienteSC.sh
View File

@ -43,7 +43,7 @@ pushTmpDir ORG.coorienteSC
loginfo "Running Blast..."
blastn -db "${BLASTDB}" -query "${DATA}" -outfmt 6 | \
awk ' \
$AwkCmd ' \
($4 > 1000) && ($3 > 70) \
($1==QUERY) && \
($2==SUBJECT) && \
@ -65,7 +65,7 @@ pushTmpDir ORG.coorienteSC
LDIFF= ($3/100.*$4) }} \
} \
END {print QUERY,SUBJECT,LSAME,LDIFF,(LSAME>LDIFF)}' | \
awk -v minlength="${MINLENGTH}" \
$AwkCmd -v minlength="${MINLENGTH}" \
' (($3>minlength) || \
($4 > minlength)) && \
($3/($4+1) > 2) && \
@ -75,7 +75,7 @@ pushTmpDir ORG.coorienteSC
{print $1,$2,$5}}' | \
sort | \
uniq -c | \
awk '($1==2) {print $2,$3,$4}'
$AwkCmd '($1==2) {print $2,$3,$4}'
loginfo "Done"
popTmpDir

2
detectors/normalize/tools/extract_refLSC.sh
View File

@ -2,7 +2,7 @@
#
awk 'function printfasta(seq) { \
gawk 'function printfasta(seq) { \
seqlen=length(seq); \
for (i=1; i <= seqlen; i+=60) \
print substr(seq,i,60); \

2
detectors/normalize/tools/extract_refSSC.sh
View File

@ -2,7 +2,7 @@
#
awk 'function printfasta(seq) { \
gawk 'function printfasta(seq) { \
seqlen=length(seq); \
for (i=1; i <= seqlen; i+=60) \
print substr(seq,i,60); \

2
detectors/normalize/tools/selectViridiplantae.sh
View File

@ -2,7 +2,7 @@
grep -A 1 ' ORGANISM' $* | \
grep -B 1 Viridiplantae | \
awk '{print $1}' | \
gawk '{print $1}' | \
grep '\.gbk' | \
sed -E 's/(^.*\.gbk).$/\1/' | \
uniq