f0527b030e6164ea3bf58446dfaef6da1d4f62dd
chmalee
  Tue Sep 14 10:41:07 2021 -0700
Fix lrg build to not depend on checked out copy of kent source, copy all scripts locally

diff --git src/hg/utils/otto/lrg/buildLrg.sh src/hg/utils/otto/lrg/buildLrg.sh
index 37b47f9..c0d97d4 100755
--- src/hg/utils/otto/lrg/buildLrg.sh
+++ src/hg/utils/otto/lrg/buildLrg.sh
@@ -5,31 +5,31 @@
 WORKDIR=$1
 mkdir -p ${WORKDIR}/${today}
 cd ${WORKDIR}/${today}
 
 for db in hg38 hg19
 do
     grc=""
     if [ $db == "hg38" ]; then
         grc="GRCh38"
     else
         grc="GRCh37"
     fi
     mkdir -p $db
     cd $db
     echo "building LRG for $db"
-    ~/kent/src/hg/utils/automation/parseLrgXml.pl $grc ../
+    ${WORKDIR}/parseLrgXml.pl $grc ../
     set +e
     genePredCheck lrgTranscriptsUnmapped.gp 2>genePred.failed
     set -e
     # use -gt 1 because one of the lines will be the "checked: X, failed: Y" line
     if [ `grep -v "LRG_7t1\|LRG_1298t1" genePred.failed | wc -l` -gt 2 ]; then
         printf "genePredCheck failed on %s. Check %s for more info\n" "${db}" "${WORKDIR}/${today}/${db}/genePred.failed"
         exit 255
     fi
     cut -f1,12 lrgTranscriptsUnmapped.gp | sort > transcript.gene.name.txt
 
     awk -F$'\t' '{printf "%s\t%s\t%s\t%s\t%s\t%s\t%s %s %s %s\n", $1,$16,$17,$18,$19,$2, $16,$18,$17,$19}' \
         lrgTranscriptsUnmapped.gp | sort \
         | join -t$'\t' - transcript.gene.name.txt \
         | awk -F$'\t' '{printf "%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s %s\n", $1,$2,$3,$4,$5,$6,$8,$7,$8}' \
         > lrgTransExtraFields.tsv
@@ -61,20 +61,26 @@
     printf "Creating lrgBigPsl.bb for %s\n" $db
     bedToBigBed -as=${WORKDIR}/bigPsl+6.as -type=bed12+20 -tab -extraIndex=name \
         lrgExtraTranscripts${db}.bigPsl.bed /hive/data/genomes/${db}/chrom.sizes lrgBigPsl.bb
 
     # The lrg table already exists as a one line table pointing to /gbdb/$db/bbi/lrg.bb, which
     # is in turn a symlink into $WORKDIR/release/$db/lrg.bb. Thus validate now before overwriting:
     oldCount=`bigBedInfo ${WORKDIR}/release/$db/lrg.bb | grep itemCount | cut -d' ' -f2`
     oldPslCount=`bigBedInfo ${WORKDIR}/release/$db/lrgBigPsl.bb | grep itemCount | cut -d' ' -f2`
     newCount=`bigBedInfo lrg.bb | grep itemCount | cut -d' ' -f2`
     newPslCount=`bigBedInfo lrgBigPsl.bb | grep itemCount | cut -d' ' -f2`
     echo LRG rowcounts: old $oldCount new: $newCount
     echo $oldCount $newCount | awk '{if (($2-$1)/$1 > 0.1) {printf "validate on LRG BigBed failed: old count: %d, new count: %d\n", $1,$2; exit 1;}}'
     echo LRG PSL rowcounts: old $oldPslCount new: $newPslCount
     echo $oldPslCount $newPslCount | awk '{if (($2-$1)/$1 > 0.1) {printf "validate on DECIPHER CNV failed: old count: %d, new count: %d\n", $1,$2; exit 1;}}'
 
+    # now archive this new release:
+    archiveDir="/usr/local/apache/htdocs-hgdownload/goldenPath/archive/${db}/lrg/${today}"
+    mkdir -p ${archiveDir}
+    cp lrg.bb ${archiveDir}
+    cp lrgBigPsl.bb ${archiveDir}
+    
     cp lrg.bb ${WORKDIR}/release/${db}/
     cp lrgBigPsl.bb ${WORKDIR}/release/${db}/
 
     cd ${WORKDIR}/${today}
 done