#!/usr/bin/env perl use strict; use warnings; use File::Basename; my $thisMachine = `uname -n`; chomp $thisMachine; if ($thisMachine ne "hgdownload") { if ($thisMachine ne "hgdownload1") { if ($thisMachine ne "hgdownload1.soe.ucsc.edu") { printf STDERR "# NOTE: This script is only used on hgdownload\n"; exit 255; } } } ############################################################################# sub startHtml() { printf '

GenArk: UCSC Genome Archive

' } ############################################################################# sub endHtml() { printf ' ' } ############################################################################# startHtml; my %expectedList = ( "VGP" => 1, "CCGP" => 1, "HPRC" => 1, "birds" => 1, "fish" => 1, "globalReference" => 1, "mammals" => 1, "primates" => 1, "vertebrate" => 1, "invertebrate" => 1, "fungi" => 1, "legacy" => 1, "plants" => 1, "viral" => 1, "bacteria" => 1, ); my %titles = ( "VGP" => "Vertebrate Genomes Project collection", "CCGP" => "The California Conservation Genomics Project", "HPRC" => "Human Pangenome Reference Consortium", "birds" => "NCBI bird genomes", "fish" => "NCBI fish genomes", "globalReference" => "Global Human Reference genomes, January 2020", "mammals" => "NCBI mammal genomes", "primates" => "NCBI primate genomes", "vertebrate" => "NCBI other vertebrate genomes", "invertebrate" => "NCBI invertebrate genomes", "fungi" => "NCBI fungi genomes", "legacy" => "NCBI genomes legacy/superseded by newer versions", "plants" => "NCBI plant genomes", "viral" => "NCBI virus genomes", "bacteria" => "NCBI bacteria genomes", "gtexAnalysis" => "Genotype-Tissue Expression (GTEx) Project analysis results track hub, V6 October 2015", "gtex" => "Genotype-Tissue Expression (GTEx) RNA-seq signal track hub, V6 October 2015", "mouseStrains" => "16 mouse strain assembly and track hub, May 2017", "neuroDiffCrispr" => "Structurally conserved primate cerebral cortex lincRNAs track hub, December 2018", ); my %otherTopLevels = ( "GCA" => 1, "GCF" => 1, "gtex" => 1, "gtexAnalysis" => 1, "mouseStrains" => 1, "neuroDiffCrispr" => 1, "UCSC_GI.assemblyHubList.txt" => 1, "index.html" => 1, ); my @orderOutHubs = ( "primates", "mammals", "birds", "fish", "vertebrate", "invertebrate", "fungi", "plants", "viral", "bacteria", "VGP", "CCGP", "HPRC", "globalReference", "mouseStrains", "legacy", ); my @orderOutTracks = ( "gtexAnalysis", "gtex", "neuroDiffCrispr", ); my %indexPage = ( "primates" => "index.html", "mammals" => "index.html", "birds" => "index.html", "fish" => "index.html", "vertebrate" => "index.html", "invertebrate" => "index.html", "fungi" => "index.html", "legacy" => "index.html", "plants" => "index.html", "viral" => "index.html", "bacteria" => "index.html", "VGP" => "index.html", "CCGP" => "index.html", "HPRC" => "index.html", "mouseStrains" => "hubIndex.html", "globalReference" => "index.html", "gtexAnalysis" => "index.html", "gtex" => "index.html", "neuroDiffCrispr" => "index.html", ); # verify all known directories and files, alert for any new ones open (FH, "ls -d /mirrordata/hubs/*|") or die "can not ls -d /mirrordata/hubs/*"; while (my $dirPath = ) { chomp $dirPath; my $fileDirName = basename($dirPath); if (! (defined($expectedList{$fileDirName}) || defined($otherTopLevels{$fileDirName})) ) { printf STDERR "# something new: %s\n", $fileDirName; } } close (FH); ### Determine genome counts: my %genomeCounts; my $genomeCount = `grep -h ^genome /mirrordata/hubs/VGP/*enomes.txt | wc -l`; chomp $genomeCount; $genomeCounts{"VGP"} = $genomeCount; my @checkList = ('primates', 'mammals', 'birds', 'fish', 'vertebrate', 'legacy', 'plants', "invertebrate", "fungi", 'viral', 'bacteria', 'CCGP', 'HPRC', 'globalReference'); foreach my $hubSet (@checkList) { $genomeCount = `grep -h ^genome /mirrordata/hubs/$hubSet/genomes.txt | wc -l`; chomp $genomeCount; $genomeCounts{$hubSet} = $genomeCount; } my $hubCount = 0; printf "\n"; printf "\n"; printf " \n"; printf " \n"; printf "\n"; # construct table foreach my $orderUp (@orderOutHubs) { printf "\n"; ++$hubCount; if ($orderUp eq "VGP") { printf " \n"; printf "\n"; printf "\n"; } if ($orderUp eq "fish") { printf " \n", $orderUp, $indexPage{$orderUp}; } else { printf " \n", $orderUp, $indexPage{$orderUp}, $orderUp; } if (defined($genomeCounts{$orderUp})) { printf " \n", $titles{$orderUp}, $genomeCounts{$orderUp}; } else { printf " \n", $titles{$orderUp}; } printf "\n"; } printf "
hub gatewaydescription
collections below are subsets of the assemblies above
fishes%s%s (%d assemblies)%s
\n"; my $totalAsmHubs = `grep -v "^#" /mirrordata/hubs/UCSC_GI.assemblyHubList.txt | wc -l`; chomp $totalAsmHubs; printf "

\n"; printf "Please note: text file listing of %d genome assembly hubs\n", $totalAsmHubs; printf "

\n"; printf "

\n"; printf "Please note, the invertebrate category contains more than just invertebrate animals. Until these clades are sorted out, with extra categories created, you will also find parasites, protozoa, and other single celled eukaryotes in the invertebrate grouping.\n"; printf "

\n"; printf "

\n"; printf "Please use the Assembly Request page to find and request GenBank assemblies that have not yet been included in the collections here.\n"; printf "

\n"; endHtml;