030eb76bec5c124428fd35626ea30609df945cbc
jcasper
  Wed Aug 23 15:44:06 2023 -0700
Slight name change to improve readability, no ticket

diff --git src/hg/utils/otto/knownGene/buildTo.sh src/hg/utils/otto/knownGene/buildTo.sh
index 0edf72a..c163074 100755
--- src/hg/utils/otto/knownGene/buildTo.sh
+++ src/hg/utils/otto/knownGene/buildTo.sh
@@ -1,29 +1,29 @@
 #!/bin/sh -ex
 
 {
 . ./buildEnv.sh
 
 # knownToLocusLink
 #hgsql --skip-column-names -e "select mrnaAcc,locusLinkId from hgFixed.refLink" $db > refToLl.txt
 hgsql --skip-column-names -e "select mrnaAcc,locusLinkId from ncbiRefSeqLink where mrnaAcc != ''" $db > refToLl.txt
 hgMapToGene -geneTableType=genePred -tempDb=$tempDb $db ncbiRefSeq knownGene knownToLocusLink -lookup=refToLl.txt
 rm refToLl.txt
 
-if test "$gtexGeneMode" != ""
+if test "$gtexGeneModel" != ""
 then
-    hgMapToGene -geneTableType=genePred $db -tempDb=$tempDb -all -type=genePred $gtexGeneMode knownGene knownToGtex
+    hgMapToGene -geneTableType=genePred $db -tempDb=$tempDb -all -type=genePred $gtexGeneModel knownGene knownToGtex
 fi
 
 # knownToEnsembl and knownToGencode${GENCODE_VERSION}
 awk '{OFS="\t"} {print $4,$4}' ucscGenes.bed | sort | uniq > knownToEnsembl.tab
 cp knownToEnsembl.tab knownToGencode${GENCODE_VERSION}.tab
 hgLoadSqlTab -notOnServer $tempDb  knownToEnsembl  $kent/src/hg/lib/knownTo.sql  knownToEnsembl.tab
 hgLoadSqlTab -notOnServer $tempDb  knownToGencode${GENCODE_VERSION}  $kent/src/hg/lib/knownTo.sql  knownToGencode${GENCODE_VERSION}.tab
 
 # make knownToLynx
 #wget "http://lynx.ci.uchicago.edu/downloads/LYNX_GENES.tab"
 #awk '{print $2}' LYNX_GENES.tab | sort > lynxExists.txt
 #hgsql -e "select geneSymbol,kgId from kgXref" --skip-column-names $tempDb | awk '{if (NF == 2) print}' | sort > geneSymbolToKgId.txt
 #join lynxExists.txt geneSymbolToKgId.txt | awk 'BEGIN {OFS="\t"} {print $2,$1}' | sort > knownToLynx.tab
 #hgLoadSqlTab -notOnServer $tempDb  knownToLynx $kent/src/hg/lib/knownTo.sql  knownToLynx.tab
 
@@ -93,26 +93,28 @@
 # the extDb{Ref} parts come from hg/hgGene/domains.c:domainsPrint()
 hgsql -Ne "select kgID, extAcc1 from $tempDb.kgXref x \
     inner join sp180404.extDbRef sp on x.spID = sp.acc \
     inner join sp180404.extDb e on sp.extDb=e.id \
     where x.spID != '' and e.val='PDB' order by kgID" \
     > $tempDb.knownToPdb.txt;
 # filter out pdbIds not found in mupit
 cat mupit-pdbids.txt | tr '[a-z]' '[A-Z]' | \
     grep -Fwf - $tempDb.knownToPdb.txt >  knownToMupit.txt;
 # check that it filtered correctly:
 # cut -f2 $db.knownToMuipit.txt | sort -u | wc -l;
 # load new table for hgGene/hgc
 hgLoadSqlTab $tempDb knownToMupit ~/kent/src/hg/lib/knownTo.sql knownToMupit.txt
 
 # make knownToNextProt
-wget "ftp://ftp.nextprot.org/pub/current_release/ac_lists/nextprot_ac_list_all.txt"
+#wget "ftp://ftp.nextprot.org/pub/current_release/ac_lists/nextprot_ac_list_all.txt"
+wget "https://download.nextprot.org/pub/current_release/ac_lists/nextprot_ac_list_all.txt"
 awk '{print $0, $0}' nextprot_ac_list_all.txt | sed 's/NX_//' | sort > displayIdToNextProt.txt
 hgsql -e "select spID,kgId from kgXref" --skip-column-names $tempDb | awk '{if (NF == 2) print}' | sort > displayIdToKgId.txt
 join displayIdToKgId.txt displayIdToNextProt.txt | awk 'BEGIN {OFS="\t"} {print $2,$3}' > knownToNextProt.tab
 hgLoadSqlTab -notOnServer $tempDb  knownToNextProt $kent/src/hg/lib/knownTo.sql  knownToNextProt.tab
 
-hgMapToGene -geneTableType=genePred -tempDb=$tempDb $db HInvGeneMrna knownGene knownToHInv
+# H-Inv is no longer built (last on hg19)
+#hgMapToGene -geneTableType=genePred -tempDb=$tempDb $db HInvGeneMrna knownGene knownToHInv
 
 
 echo "BuildKnownTo successfully finished"
 } > doKnownTo.log < /dev/null 2>&1