b98ba30fa69bf7ff12c58db12776589373e24e95 max Mon Nov 4 05:04:51 2019 -0800 splitting lovd track into composite with two subtracks, refs #24408 diff --git src/hg/utils/otto/lovd/checkAndLoad.sh src/hg/utils/otto/lovd/checkAndLoad.sh index 740b02f..7440d28 100755 --- src/hg/utils/otto/lovd/checkAndLoad.sh +++ src/hg/utils/otto/lovd/checkAndLoad.sh @@ -7,51 +7,54 @@ WORKDIR="/hive/data/outside/otto/lovd" TAWK=/cluster/bin/scripts/tawk KENTBIN=/cluster/bin/x86_64/ if [ ! -d "${WORKDIR}" ]; then echo "ERROR in lovd release, Can not find the directory: ${WORKDIR}" exit 255 fi cd "${WORKDIR}" today=`date +%F` cd $today # count columns and make sure new files have 6 columns new19Cc=`cat lovd.hg19.bed | $TAWK '(NF!=6)' | wc -l` -new18Cc=`cat lovd.hg18.bed | $TAWK '(NF!=6)' | wc -l` +#new18Cc=`cat lovd.hg18.bed | $TAWK '(NF!=6)' | wc -l` if [ "$new19Cc" -ne "0" ]; then echo LVOD hg19 $today: found rows with not six columns, quitting exit 255 fi -if [ "$new18Cc" -ne "0" ]; then - echo LVOD hg18 $today: found rows with not six columns, quitting - exit 255 -fi + +#if [ "$new18Cc" -ne "0" ]; then +# echo LVOD hg18 $today: found rows with not six columns, quitting +# exit 255 +#fi # compare old and new line counts and abort if no increase old19Lc=`$KENTBIN/hgsql hg19 -e "SELECT COUNT(*) from lovd" -NB` new19Lc=`wc -l lovd.hg19.bed | cut -d' ' -f1 ` -old18Lc=`$KENTBIN/hgsql hg18 -e "SELECT COUNT(*) from lovd" -NB` -new18Lc=`wc -l lovd.hg18.bed | cut -d' ' -f1 ` +#old18Lc=`$KENTBIN/hgsql hg18 -e "SELECT COUNT(*) from lovd" -NB` +#new18Lc=`wc -l lovd.hg18.bed | cut -d' ' -f1 ` echo hg19 rowcount: old $old19Lc new: $new19Lc -echo hg18 rowcount: old $old18Lc new: $new18Lc +#echo hg18 rowcount: old $old18Lc new: $new18Lc if [ "$new19Lc" -eq "$old19Lc" ]; then echo LVOD hg19: rowcount for $today is equal to old rowcount in mysql, quitting exit 0 fi # commenting out, as LOVD count will go down in the future, because they have a # db cleanup procedure in place now -#if [ "$new19Lc" -lt "$old19Lc" ]; then - #echo LVOD hg19: rowcount for $today is smaller to old rowcount in mysql, quitting - #exit 255 -#fi +if [ "$new19Lc" -lt "$old19Lc" ]; then + echo LOVD hg19: rowcount for $today is smaller to old rowcount in mysql, quitting + exit 255 +fi # bedDetail4.sql was generated like this: # egrep -v 'score|strand|thick|reserved|block|chromStarts' /cluster/home/max/kent/src/hg/lib/bedDetail.sql > bedDetail4.sql # need to use bedClip as current files include invalid coords which LOVD won't fix. -$KENTBIN/bedClip lovd.hg19.bed /cluster/data/hg19/chrom.sizes stdout | $KENTBIN/hgLoadBed hg19 lovd stdin -tab -sqlTable=../bedDetail4.sql -renameSqlTable -noBin -$KENTBIN/bedClip lovd.hg18.bed /cluster/data/hg18/chrom.sizes stdout | $KENTBIN/hgLoadBed hg18 lovd stdin -tab -sqlTable=../bedDetail4.sql -renameSqlTable -noBin +$KENTBIN/bedClip lovd.hg19.bed /cluster/data/hg19/chrom.sizes lovd.hg19.clipped.bed +cat lovd.hg19.clipped.bed | awk '(($3-$2)<=100)' | $KENTBIN/hgLoadBed hg19 lovdShort stdin -tab -sqlTable=../bedDetail4.sql -renameSqlTable -noBin +cat lovd.hg19.clipped.bed | awk '(($3-$2)>100)' | $KENTBIN/hgLoadBed hg19 lovdLong stdin -tab -sqlTable=../bedDetail4.sql -renameSqlTable -noBin +#$KENTBIN/bedClip lovd.hg18.bed /cluster/data/hg18/chrom.sizes stdout | $KENTBIN/hgLoadBed hg18 lovd stdin -tab -sqlTable=../bedDetail4.sql -renameSqlTable -noBin