22a9ec8b34dd60a14257bdef408f21c03db601b4
hiram
  Mon Feb 17 16:06:38 2020 -0800
now using the shorter assembly name for genome identifier and fix some common names refs #23891

diff --git src/hg/makeDb/doc/asmHubs/mkGenomes.pl src/hg/makeDb/doc/asmHubs/mkGenomes.pl
index e8c2acb..4ab17f0 100755
--- src/hg/makeDb/doc/asmHubs/mkGenomes.pl
+++ src/hg/makeDb/doc/asmHubs/mkGenomes.pl
@@ -13,93 +13,95 @@
 my $Name = shift;
 my $asmHubName = shift;
 
 my %betterName;	# key is asmId, value is common name
 my $srcDocDir = "${asmHubName}AsmHub";
 my $buildDir = "/hive/data/genomes/asmHubs/refseqBuild";
 my $destDir = "/hive/data/genomes/asmHubs/$asmHubName";
 
 my $home = $ENV{'HOME'};
 my $toolsDir = "$home/kent/src/hg/makeDb/doc/asmHubs";
 my $commonNameList = "$asmHubName.asmId.commonName.tsv";
 my $commonNameOrder = "$asmHubName.commonName.asmId.orderList.tsv";
 
 open (FH, "<$toolsDir/${commonNameList}") or die "can not read $toolsDir/${commonNameList}";
 while (my $line = <FH>) {
+  next if ($line =~ m/^#/);
   chomp $line;
   my ($asmId, $name) = split('\t', $line);
   $betterName{$asmId} = $name;
 }
 close (FH);
 
 my @orderList;	# asmId of the assemblies in order from the *.list files
 # the order to read the different .list files:
 my $assemblyCount = 0;
 
 open (FH, "<$toolsDir/${commonNameOrder}") or die "can not read ${commonNameOrder}";
 while (my $line = <FH>) {
+  next if ($line =~ m/^#/);
   chomp $line;
   my ($commonName, $asmId) = split('\t', $line);
   push @orderList, $asmId;
   ++$assemblyCount;
 }
 close (FH);
 
 my $orderKey = 1;
 foreach my $asmId (reverse(@orderList)) {
   my ($gcPrefix, $accession, undef) = split('_', $asmId);
   my $accessionId = sprintf("%s_%s", $gcPrefix, $accession);
   my $accessionDir = substr($asmId, 0 ,3);
   $accessionDir .= "/" . substr($asmId, 4 ,3);
   $accessionDir .= "/" . substr($asmId, 7 ,3);
   $accessionDir .= "/" . substr($asmId, 10 ,3);
   my $buildDir = "/hive/data/genomes/asmHubs/refseqBuild/" . substr($asmId, 0 ,3);
   $buildDir .= "/" . substr($asmId, 4 ,3);
   $buildDir .= "/" . substr($asmId, 7 ,3);
   $buildDir .= "/" . substr($asmId, 10 ,3);
   $buildDir .= "/" . $asmId;
   my $asmReport="$buildDir/download/${asmId}_assembly_report.txt";
   next if (! -s $asmReport);
 printf STDERR "# %03d genomes.txt %s/%s\n", $orderKey, $accessionDir, $accessionId;
   my $descr=`grep -i "organism name:" $asmReport | head -1 | sed -e 's#.*organism name: *##i; s# (.*\$##;'`;
   chomp $descr;
   my $orgName=`grep -i "organism name:" $asmReport | head -1 | sed -e 's#.* name: .* (##; s#).*##;'`;
   chomp $orgName;
   $orgName = $betterName{$asmId} if (exists($betterName{$asmId}));
 
-  printf "genome %s\n", $asmId;
-  printf "trackDb %s/%s/trackDb.txt\n", $accessionDir, $accessionId;
+  printf "genome %s\n", $accessionId;
+  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 "twoBitPath ../%s/%s/%s.2bit\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", $orderKey++;
   printf "scientificName %s\n", $descr;
-  printf "htmlPath %s/%s/html/%s.description.html\n", $accessionDir, $accessionId, $asmId;
+  printf "htmlPath ../%s/%s/html/%s.description.html\n", $accessionDir, $accessionId, $asmId;
   printf "\n";
   my $localGenomesFile = "$buildDir/${asmId}.genomes.txt";
   my $localOrderKey;
   open (GF, ">$localGenomesFile") or die "can not write to $localGenomesFile";
   printf GF "genome %s\n", $accessionId;
   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 "organism %s\n", $descr;
   printf GF "defaultPos %s\n", $defPos;
   printf GF "orderKey %d\n", $localOrderKey++;
   printf GF "scientificName %s\n", $descr;
   printf GF "htmlPath html/%s.description.html\n", $asmId;
   close (GF);