430076e29a82930d89c6290946436fd6781e18f2 hiram Thu Mar 17 10:05:04 2022 -0700 adding bpt indexes for 2bit files and correct copy of 2bit file and gfidx files and fixup group titles refs #28961 diff --git src/hg/makeDb/doc/asmHubs/mkGenomes.pl src/hg/makeDb/doc/asmHubs/mkGenomes.pl index 122303d..b497bb0 100755 --- src/hg/makeDb/doc/asmHubs/mkGenomes.pl +++ src/hg/makeDb/doc/asmHubs/mkGenomes.pl @@ -12,59 +12,61 @@ printf STDERR "the name list is found in \$HOME/kent/src/hg/makeDb/doc/asmHubs/\n"; printf STDERR "\nthe two columns are 1: asmId (accessionId_assemblyName)\n"; printf STDERR "column 2: common name for species, columns separated by tab\n"; 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) = @_; 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; 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; @@ -153,30 +155,31 @@ chomp $taxId; 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 $orgName; if (defined($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; 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 $defPos; } @@ -210,90 +213,50 @@ open (DL, ">$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); 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; 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/${asmId}.groups.txt"; open (GR, ">$localGroups") or die "can not write to $localGroups"; - print GR <<_EOF_ -name user -label Custom Tracks -priority 1 -defaultIsClosed 1 - -name map -label Mapping and Sequencing -priority 2 -defaultIsClosed 0 - -name genes -label Genes and Gene Predictions -priority 3 -defaultIsClosed 0 - -name rna -label mRNA and EST -priority 4 -defaultIsClosed 0 - -name regulation -label Regulation -priority 5 -defaultIsClosed 0 - -name compGeno -label Comparative Genomics -priority 6 -defaultIsClosed 0 - -name varRep -label Variation -priority 7 -defaultIsClosed 0 - -name x -label Experimental -priority 10 -defaultIsClosed 1 -_EOF_ - ; + print GR "%s", $groupsTxt; close (GR); } __END__ description Mastacembelus armatus twoBitPath GCA_900324485.2_fMasArm1.2/trackData/addMask/GCA_900324485.2_fMasArm1.2.masked.2bit organism Zig-Zag eel defaultPos LR535842.1:14552035-14572034 orderKey 1 scientificName Mastacembelus armatus htmlPath GCA_900324485.2_fMasArm1.2/html/GCA_900324485.2_fMasArm1.2.description.html # head -25 GCA_002180035.3_HG00514_prelim_3.0_assembly_report.txt