351ed2fc24a13b86cf2a1fc9695309e2c0da9de5
lrnassar
  Wed Jun 3 09:33:28 2026 -0700
Make otto fetch failures loud and clinGen silent on no-update. refs #37697

Several otto jobs used 'wget -q'/'curl -s' under 'set -e', so a broken download
aborted with no output and cron (which mails on output, not exit code) sent nothing
-- the mode by which mastermind failed silently for months.

- mastermind, malacards: add an ERR trap that echoes a failure message; switch
mastermind's 'curl -s' to 'curl -sf' so an HTTP error fails loudly instead of
writing a challenge page into the release file.
- geneReviews: add an ERR trap (kept silent on no-update via the wrapper's mail -E).
- All three use 'set -eE' (errtrace) so the ERR trap also fires for failures inside
functions/subshells, not just top-level commands.
- clinGen: capture each sub-build's exit status instead of swallowing it under
'set +e', and report which sub-build failed. Also drop the routine 'No ... update'
echoes and the unconditional 'ClinGen update done.' line, and switch the wrapper to
'mail -E', so clinGen is now silent on no-update and only emails on a real update
or a failure.

diff --git src/hg/utils/otto/clinGen/makeDosage.sh src/hg/utils/otto/clinGen/makeDosage.sh
index e1f8e377167..a4cb947c293 100755
--- src/hg/utils/otto/clinGen/makeDosage.sh
+++ src/hg/utils/otto/clinGen/makeDosage.sh
@@ -1,109 +1,107 @@
 #!/bin/bash
 
 # downloads and builds the curated ClinGen Curated CNVS track
 # assumes running in the build directory:
 # /hive/data/outside/otto/clinGen/
 set -beEu -o pipefail
 
 WORKDIR=$1
 mkdir -p ${WORKDIR}/clinGenDosage
 cd ${WORKDIR}/clinGenDosage
 
 echo "user anonymous
 ls ClinGen*curation_list*
 bye" > ftp.dosage.cmds
 
 if [ -e release.list ]
 then
     mv release.list prev.release.list
 fi
 touch prev.release.list
 rm -f release.list
 
 # connect and list a directory, result to file: ls.check
 ftp -n -v -i ftp.clinicalgenome.org 2>&1 < ftp.dosage.cmds &> ls.check
 grep "haplo\|triplo\|curation" ls.check | sort > release.list || echo "Error - no bed files found"
 
 # see if anything is changing, if so, notify, download, and build
 diff prev.release.list release.list > release.diff || true
 count=`wc -l release.diff | cut -d' ' -f1`
 if [ "${count}" -gt 1 ]
 then
     echo "New ClinGen Dosage update"
     today=`date +%F`
     mkdir -p ${today}/{download,output}
     cd ${today}/download
     for db in hg19 hg38
     do
         grc=""
         if [ ${db} == "hg19" ]
         then
             grc="GRCh37"
             wget -N -q "ftp://ftp.clinicalgenome.org/ClinGen_region_curation_list_${grc}.tsv"
             wget -N -q "ftp://ftp.clinicalgenome.org/ClinGen_gene_curation_list_${grc}.tsv"
         elif [ ${db} == "hg38" ]
         then
             grc="GRCh38"
             wget -N -q "ftp://ftp.clinicalgenome.org/ClinGen_region_curation_list_${grc}.tsv"
             wget -N -q "ftp://ftp.clinicalgenome.org/ClinGen_gene_curation_list_${grc}.tsv"
         fi
         echo $grc
         # the process script creates a haplo.bed and triplo.bed file:
         ../../../processClinGenDosage.py ClinGen_region_curation_list_${grc}.tsv ClinGen_gene_curation_list_${grc}.tsv ../output/${db}
         sort -k1,1 -k2,2n ../output/${db}.haplo.bed > ../output/${db}.clinGenHaplo.bed
         sort -k1,1 -k2,2n ../output/${db}.triplo.bed > ../output/${db}.clinGenTriplo.bed
 
         # validate new release by diffing against old release:
         bigBedToBed ${WORKDIR}/release/${db}/clinGenHaplo.bb ${db}.old.haplo.bed
         bigBedToBed ${WORKDIR}/release/${db}/clinGenTriplo.bb ${db}.old.triplo.bed
         oldHaploLc=$(wc -l ${db}.old.haplo.bed | cut -d' ' -f1)
         newHaploLc=$(wc -l ../output/${db}.clinGenHaplo.bed | cut -d' ' -f1)
         # diff returns 1 when it finds a difference, which causes the script to silently exit. Add the
         # '|| true' to prevent that behavior since we want the number of diffs
         diffCountHaplo=$( (diff ${db}.old.haplo.bed ../output/${db}.clinGenHaplo.bed --speed-large-files || true) | grep -c '^[\>\<]')
         diffPerc=$(echo $oldHaploLc $newHaploLc | awk '{diff=(($2-$1)/$1); printf "%0.2f", diff}')
         shouldError=$(echo $diffPerc | awk '{if ($1 > 0.1 || $1 < -0.1) {print 0} else {print 1}}')
         if [[ $shouldError -eq 0 ]]; then
             printf "validate on %s ClinGen Haplo failed: old count: %d, new count: %d, difference: %0.2f\n" $db $oldHaploLc $newHaploLc $diffPerc
             exit 1
         fi
         diffPerc=$(echo $newHaploLc $diffCountHaplo | awk '{diff=$2/$1; printf "%0.2f", diff}')
         shouldError=$(echo $diffPerc | awk '{if ($1 > 0.1 || $1 < -0.1) {print 0} else {print 1}}')
         if [[ $shouldError -eq 0 ]]; then
             printf "validate on %s ClinGen Haplo failed with too many differences to old version: %d lines changed, new count: %d, difference: %0.2f\n" $db $diffCountHaplo $newHaploLc $diffPerc
             exit 1
         fi
         oldTriploLc=$(wc -l ${db}.old.triplo.bed | cut -d' ' -f1)
         newTriploLc=$(wc -l ../output/${db}.clinGenTriplo.bed | cut -d' ' -f1)
         # diff returns 1 when it finds a difference, which causes the script to silently exit. Add the
         # '|| true' to prevent that behavior since we want the number of diffs
         diffCountTriplo=$( (diff ${db}.old.triplo.bed ../output/${db}.clinGenTriplo.bed --speed-large-files || true) | grep -c '^[\>\<]')
         diffPerc=$(echo $oldTriploLc $newTriploLc | awk '{diff=(($2-$1)/$1); printf "%0.2f", diff}')
         shouldError=$(echo $diffPerc | awk '{if ($1 > 0.1 || $1 < -0.1) {print 0} else {print 1}}')
         if [[ $shouldError -eq 0 ]]; then
             printf "validate on %s ClinGen Triplo failed: old count: %d, new count: %d, difference: %0.2f\n" $db $oldTriploLc $newTriploLc $diffPerc
             exit 1
         fi
         diffPerc=$(echo $newTriploLc $diffCountTriplo | awk '{diff=$2/$1; printf "%0.2f", diff}')
         shouldError=$(echo $diffPerc | awk '{if ($1 > 0.1 || $1 < -0.1) {print 0} else {print 1}}')
         if [[ $shouldError -eq 0 ]]; then
             printf "validate on %s ClinGen Triplo failed with too many differences to old version: %d lines changed, new count: %d, difference: %0.2f\n" $db $diffCountTriplo $newTriploLc $diffPerc
             exit 1
         fi
 
         bedToBigBed -type=bed9+17 -as=../../clinGenDosageHaplo.as -tab ../output/${db}.clinGenHaplo.bed /hive/data/genomes/${db}/chrom.sizes ../output/${db}.clinGenHaplo.bb
         bedToBigBed -type=bed9+17 -as=../../clinGenDosageTriplo.as -tab ../output/${db}.clinGenTriplo.bed /hive/data/genomes/${db}/chrom.sizes ../output/${db}.clinGenTriplo.bb
         cp ../output/${db}.clinGenHaplo.bb ${WORKDIR}/release/${db}/clinGenHaplo.bb
         cp ../output/${db}.clinGenTriplo.bb ${WORKDIR}/release/${db}/clinGenTriplo.bb
 
         # ClinGen stamps the curation list with its own "results as of" date (line 2,
         # e.g. "#28 Apr,2026"). Expose it via dataVersion (shared by the Haplo and Triplo
         # subtracks) so users can tie the browser data to a specific ClinGen dosage
         # release rather than our otto run date.
         dosageDate=$(sed -n 2p ClinGen_region_curation_list_${grc}.tsv | tr -d '#\r')
         printf 'ClinGen Dosage Sensitivity Map, %s\n' "$dosageDate" > ${WORKDIR}/release/${db}/clinGenDosageVersion.txt
     done
     cd ../..
-else
-    echo "No ClinGen CNV update"
 fi