9f4b0740f6d6b421da7b2c8d346b6393f4ce95a9
hiram
  Fri Mar 25 13:52:30 2022 -0700
now using a bigBed file for chromAlias function refs #27228

diff --git src/hg/makeDb/doc/asmHubs/mkGenomes.pl src/hg/makeDb/doc/asmHubs/mkGenomes.pl
index 3e04174..95756fa 100755
--- src/hg/makeDb/doc/asmHubs/mkGenomes.pl
+++ src/hg/makeDb/doc/asmHubs/mkGenomes.pl
@@ -15,60 +15,64 @@
   printf STDERR "result will write a local asmId.genomes.txt file for each hub\n";
   printf STDERR "and a local asmId.hub.txt file for each hub\n";
   printf STDERR "and a local asmId.groups.txt file for each hub\n";
   printf STDERR "and the output to stdout will be the overall genomes.txt\n";
   printf STDERR "index file for all genomes in the given list\n";
   exit 255;
 }
 
 my $downloadHost = "hgwdev";
 my @blatHosts = qw( dynablat-01 dynablat-01 );
 my @blatPorts = qw( 4040 4040 );
 my $blatHostDomain = ".soe.ucsc.edu";
 my $groupsTxt = `cat ~/kent/src/hg/makeDb/doc/asmHubs/groups.txt`;
 
 ################### writing out hub.txt file, twice ##########################
-sub singleFileHub($$$$$$$$$$) {
-  my ($fh1, $fh2, $accessionId, $orgName, $descr, $asmId, $defPos, $taxId, $trackDb, $accessionDir) = @_;
+sub singleFileHub($$$$$$$$$$$) {
+  my ($fh1, $fh2, $accessionId, $orgName, $descr, $asmId, $defPos, $taxId, $trackDb, $accessionDir, $buildDir) = @_;
   my @fhN;
   push @fhN, $fh1;
   push @fhN, $fh2;
 
   my $fileCount = 0;
   my @tdbLines;
   open (TD, "<$trackDb") or die "can not read trackDb: $trackDb";
   while (my $tdbLine = <TD>) {
      chomp $tdbLine;
      push @tdbLines, $tdbLine;
   }
   close (TD);
   foreach my $fh (@fhN) {
     printf $fh "hub %s genome assembly\n", $accessionId;
     printf $fh "shortLabel %s\n", $orgName;
     printf $fh "longLabel %s/%s/%s genome assembly\n", $orgName, $descr, $asmId;
     printf $fh "useOneFile on\n";
     printf $fh "email hclawson\@ucsc.edu\n";
     printf $fh "descriptionUrl html/%s.description.html\n", $asmId;
     printf $fh "\n";
     printf $fh "genome %s\n", $accessionId;
     printf $fh "taxId %s\n", $taxId if (length($taxId) > 1);
     printf $fh "groups groups.txt\n";
     printf $fh "description %s\n", $orgName;
     printf $fh "twoBitPath %s.2bit\n", $accessionId;
     printf $fh "twoBitBptUrl %s.2bit.bpt\n", $accessionId;
     printf $fh "chromSizes %s.chrom.sizes.txt\n", $accessionId;
+    if ( -s "${buildDir}/${asmId}.chromAlias.bb" ) {
+      printf $fh "chromAliasBb %s.chromAlias.bb\n", $accessionId;
+    } else {
       printf $fh "chromAlias %s.chromAlias.txt\n", $accessionId;
+    }
     printf $fh "organism %s\n", $descr;
     printf $fh "defaultPos %s\n", $defPos;
     printf $fh "scientificName %s\n", $descr;
     printf $fh "htmlPath html/%s.description.html\n", $asmId;
     # until blat server host is ready for hgdownload, avoid these lines
     if ($blatHosts[$fileCount] ne $downloadHost) {
       printf $fh "blat %s%s %s dynamic $accessionDir/$accessionId\n", $blatHosts[$fileCount], $blatHostDomain, $blatPorts[$fileCount];
       printf $fh "transBlat %s%s %s dynamic $accessionDir/$accessionId\n", $blatHosts[$fileCount], $blatHostDomain, $blatPorts[$fileCount];
       printf $fh "isPcr %s%s %s dynamic $accessionDir/$accessionId\n", $blatHosts[$fileCount], $blatHostDomain, $blatPorts[$fileCount];
     }
     printf $fh "\n";
     foreach my $tdbLine (@tdbLines) {
       printf $fh "%s\n", $tdbLine;
     }
     ++$fileCount;
@@ -126,30 +130,34 @@
   $accessionDir .= "/" . substr($asmId, 7 ,3);
   $accessionDir .= "/" . substr($asmId, 10 ,3);
   my $buildDir = "/hive/data/genomes/asmHubs/refseqBuild/$accessionDir/$asmId";
   my $destDir = "/hive/data/genomes/asmHubs/$accessionDir/$accessionId";
   if ($gcPrefix eq "GCA") {
      $buildDir = "/hive/data/genomes/asmHubs/genbankBuild/$accessionDir/$asmId";
   }
   if ( ! -s "${buildDir}/${asmId}.chrom.sizes" ) {
     printf STDERR "# ERROR: missing ${asmId}.chrom.sizes in\n# ${buildDir}\n";
     next;
   }
   if ( ! -s "${buildDir}/${asmId}.chromAlias.txt" ) {
     printf STDERR "# ERROR: missing ${asmId}.chromAlias.txt in\n# ${buildDir}\n";
     next;
   }
+  if ( ! -s "${buildDir}/${asmId}.chromAlias.bb" ) {
+    printf STDERR "# ERROR: missing ${asmId}.chromAlias.bb in\n# ${buildDir}\n";
+    next;
+  }
   my $asmReport="$buildDir/download/${asmId}_assembly_report.txt";
   my $trackDb = "$buildDir/$asmId.trackDb.txt";
   if ( ! -s "${trackDb}" ) {
     printf STDERR "# %03d not built yet: %s\n", $orderKey, $asmId;
     printf STDERR "# '%s'\n", $trackDb;
     next;
   }
   if ( ! -s "${asmReport}" ) {
     printf STDERR "# %03d missing assembly_report: %s\n", $orderKey, $asmId;
     next;
   }
   ++$buildDone;
 printf STDERR "# %03d genomes.txt %s/%s\n", $buildDone, $accessionDir, $accessionId;
   my $taxId=`grep -i "taxid:" $asmReport | head -1 | awk '{printf \$(NF)}' | tr -d \$'\\r'`;
   chomp $taxId;
@@ -157,31 +165,36 @@
   chomp $descr;
   my $orgName=`grep -i "organism name:" $asmReport | head -1 | sed -e 's#.* name: .* (##; s#).*##;'`;
   chomp $orgName;
   if (defined($commonName{$asmId})) {
      $orgName = $commonName{$asmId};
   }
 
   printf "genome %s\n", $accessionId;
   printf "taxId %s\n", $taxId if (length($taxId) > 1);
   printf "trackDb ../%s/%s/trackDb.txt\n", $accessionDir, $accessionId;
   printf "groups groups.txt\n";
   printf "description %s\n", $orgName;
   printf "twoBitPath ../%s/%s/%s.2bit\n", $accessionDir, $accessionId, $accessionId;
   printf "twoBitBptUrl ../%s/%s/%s.2bit.bpt\n", $accessionDir, $accessionId, $accessionId;
   printf "chromSizes ../%s/%s/%s.chrom.sizes.txt\n", $accessionDir, $accessionId, $accessionId;
+
+  if ( -s "${buildDir}/${asmId}.chromAlias.bb" ) {
+    printf "chromAliasBb ../%s/%s/%s.chromAlias.bb\n", $accessionDir, $accessionId, $accessionId;
+  } else {
     printf "chromAlias ../%s/%s/%s.chromAlias.txt\n", $accessionDir, $accessionId, $accessionId;
+  }
   printf "organism %s\n", $descr;
   my $chrName=`head -1 $buildDir/$asmId.chrom.sizes | awk '{print \$1}'`;
   chomp $chrName;
   my $bigChrom=`head -1 $buildDir/$asmId.chrom.sizes | awk '{print \$NF}'`;
   chomp $bigChrom;
   my $oneThird = int($bigChrom/3);
   my $tenK = $oneThird + 10000;
   $tenK = $bigChrom if ($tenK > $bigChrom);
   my $defPos="${chrName}:${oneThird}-${tenK}";
   if ( -s "$asmId/defaultPos.txt" ) {
     $defPos=`cat "$asmId/defaultPos.txt"`;
     chomp $defPos;
   }
   printf "defaultPos %s\n", $defPos;
   printf "orderKey %d\n", $buildDone;
@@ -203,43 +216,47 @@
   printf HT "hub %s genome assembly\n", $accessionId;
   printf HT "shortLabel %s\n", $orgName;
   printf HT "longLabel %s/%s/%s genome assembly\n", $orgName, $descr, $asmId;
   printf HT "genomesFile genomes.txt\n";
   printf HT "email hclawson\@ucsc.edu\n";
   printf HT "descriptionUrl html/%s.description.html\n", $asmId;
   close (HT);
 
   # try creating single file hub.txt, one for hgwdev, one for hgdownload
   my $downloadHubTxt = "$buildDir/${asmId}.download.hub.txt";
   open (DL, ">$downloadHubTxt") or die "can not write to $downloadHubTxt";
   $localHubTxt = "$buildDir/${asmId}.singleFile.hub.txt";
   open (HT, ">$localHubTxt") or die "can not write to $localHubTxt";
 
   singleFileHub(\*HT, \*DL, $accessionId, $orgName, $descr, $asmId,
-	$defPos, $taxId, $trackDb, $accessionDir);
+	$defPos, $taxId, $trackDb, $accessionDir, $buildDir);
 
   my $localGenomesFile = "$buildDir/${asmId}.genomes.txt";
   open (GF, ">$localGenomesFile") or die "can not write to $localGenomesFile";
   printf GF "genome %s\n", $accessionId;
   printf GF "taxId %s\n", $taxId if (length($taxId) > 1);
   printf GF "trackDb trackDb.txt\n";
   printf GF "groups groups.txt\n";
   printf GF "description %s\n", $orgName;
   printf GF "twoBitPath %s.2bit\n", $accessionId;
   printf GF "twoBitBptUrl %s.2bit.bpt\n", $accessionId;
   printf GF "chromSizes %s.chrom.sizes.txt\n", $accessionId;
+  if ( -s "${buildDir}/${asmId}.chromAlias.bb" ) {
+    printf GF "chromAliasBb %s.chromAlias.bb\n", $accessionId;
+  } else {
     printf GF "chromAlias %s.chromAlias.txt\n", $accessionId;
+  }
   printf GF "organism %s\n", $descr;
   printf GF "defaultPos %s\n", $defPos;
   printf GF "scientificName %s\n", $descr;
   printf GF "htmlPath html/%s.description.html\n", $asmId;
   # until blat server host is ready for hgdownload, avoid these lines
   if ($blatHost ne $downloadHost) {
     if ( -s "${destDir}/$accessionId.trans.gfidx" ) {
       printf GF "blat $blatHost$blatHostDomain $blatPort dynamic $accessionDir/$accessionId\n";
       printf GF "transBlat $blatHost$blatHostDomain $blatPort dynamic $accessionDir/$accessionId\n";
      printf GF "isPcr $blatHost$blatHostDomain $blatPort dynamic $accessionDir/$accessionId\n";
     }
   }
   close (GF);
 
   my $localGroups = "$buildDir/${asmId}.groups.txt";