c8828612e3aba9e6c3072768129de10abe337878 hiram Fri May 29 11:29:04 2020 -0700 now allow orderList to be specifically stated and begin common set of html functions refs #23734 diff --git src/hg/makeDb/doc/asmHubs/mkHubIndex.pl src/hg/makeDb/doc/asmHubs/mkHubIndex.pl index c88c340..e2ee199 100755 --- src/hg/makeDb/doc/asmHubs/mkHubIndex.pl +++ src/hg/makeDb/doc/asmHubs/mkHubIndex.pl @@ -1,361 +1,344 @@ #!/usr/bin/env perl +# +# mkHubIndex.pl - construct index.html page for a set of assemblies in a hub +# use strict; use warnings; +use FindBin qw($Bin); +use lib "$Bin"; +use commonHtml; my $argc = scalar(@ARGV); if ($argc != 4) { printf STDERR "mkHubIndex.pl Name asmName defaultAsmId [two column name list] > index.html\n"; printf STDERR "e.g.: mkHubIndex Primates primates GCF_000001405.39_GRCh38.p13 primates.commonName.asmId.orderList.tsv\n"; 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 "The result prints to stdout the index.html page for this set of assemblies\n"; exit 255; } +my $home = $ENV{'HOME'}; +my $toolsDir = "$home/kent/src/hg/makeDb/doc/asmHubs"; + my $Name = shift; my $asmHubName = shift; my $defaultAssembly = shift; -my $commonNameOrder = shift; - -printf STDERR "# mkHubIndex %s %s %s %s\n", $Name, $asmHubName, $defaultAssembly, $commonNameOrder; +my $inputList = shift; +my $orderList = $inputList; +if ( ! -s "$orderList" ) { + $orderList = $toolsDir/$inputList; +} -my $home = $ENV{'HOME'}; -my $toolsDir = "$home/kent/src/hg/makeDb/doc/asmHubs"; +printf STDERR "# mkHubIndex %s %s %s %s\n", $Name, $asmHubName, $defaultAssembly, $orderList; my $vgpIndex = 0; $vgpIndex = 1 if ($Name =~ m/vgp/i); my %vgpClass; # key is asmId, value is taxon 'class' as set by VGP project if ($vgpIndex) { - my $vgpClass = "$home/kent/src/hg/makeDb/doc/vgpAsmHub/vgp.taxId,asmId.class.txt"; + my $vgpClass = "$home/kent/src/hg/makeDb/doc/vgpAsmHub/vgp.taxId.asmId.class.txt"; open (FH, "<$vgpClass") or die "can not read $vgpClass"; while (my $line = ) { my ($taxId, $asmId, $class) = split('\t', $line); $vgpClass{$asmId} = $class; } close (FH); } my @orderList; # asmId of the assemblies in order from the *.list files # the order to read the different .list files: my $assemblyCount = 0; my %commonName; # key is asmId, value is a common name, perhaps more appropriate # than found in assembly_report file ############################################################################## # from Perl Cookbook Recipe 2.17, print out large numbers with comma delimiters: ############################################################################## sub commify($) { my $text = reverse $_[0]; $text =~ s/(\d\d\d)(?=\d)(?!\d*\.)/$1,/g; return scalar reverse $text } ############################################################################## ### start the HTML output ############################################################################## sub startHtml() { my $timeStamp = `date "+%F"`; chomp $timeStamp; # my $subSetMessage = "subset of $asmHubName only"; if ($asmHubName eq "vertebrate") { $subSetMessage = "subset of other ${asmHubName}s only"; } if ($vgpIndex) { + my $vgpSubset = "(set of primary assemblies)"; + if ($orderList =~ m/vgp.alternate/) { + $vgpSubset = "(set of alternate/haplotype assemblies)"; + } elsif ($orderList =~ m/vgp.trio/) { + $vgpSubset = "(set of trio assemblies, maternal/paternal)"; + } elsif ($orderList =~ m/vgp.legacy/) { + $vgpSubset = "(set of legacy/superseded assemblies)"; + } print <<"END"

VGP - Vertebrate Genomes Project assembly hub

VGP logo

This assembly hub contains assemblies released by the -Vertebrate Genomes Project. +Vertebrate Genomes Project. $vgpSubset

END } else { print <<"END"

$Name Genomes assembly hubs

Assemblies from NCBI/Genbank/Refseq sources, $subSetMessage.

END } print <<"END"

How to view the hub

Options:

  1. The links to the genome browser in the table below will attach that one specific assembly to the genome browser. This is most likely what you want.
  2. Instead, you can attach the entire set of assemblies as one group to the genome browser with the following links depending upon which of our mirror site browsers you prefer to use:
  3. To manually attach all the assemblies in this hub to genome browsers that are not one of the three UCSC mirror sites:
    1. From the blue navigation bar, go to My Data -> Track Hubs
    2. Then select the My Hubs tab and enter this URL into the textbox:
      https://hgdownload.soe.ucsc.edu/hubs/$asmHubName/hub.txt
    3. Once you have added the URL to the entry form, press the Add Hub button to add the hub.

After adding the hub, you will be redirected to the gateway page. The genome assemblies can be selected from the ${Name} Hub Assembly dropdown menu. Instead of adding all the assemblies in one collected group, use the individual view in browser in the table below.

See also: assembly statistics, track statistics <== additional information for these assemblies.


Data resource links

NOTE: Click on the column headers to sort the table by that column
The common name/view in browser will attach only that single assembly to the genome browser.
The scientific name/and data download link provides access to the files for that one assembly hub.
The class/VGP link provides access to the VGP GenomeArk page for that genome
The other links provide access to NCBI resources for these assemblies.

END } # sub startHtml() ############################################################################## ### start the table output ############################################################################## sub startTable() { print ' '; if ($vgpIndex) { printf "\n"; } print "\n"; } # sub startTable() ############################################################################## ### end the table output ############################################################################## sub endTable() { print <<"END"
count common name and
view in browser		 scientific name
and data download		 NCBI assembly bioSamplebioProject assembly date,
source link		class
VGP link
END } # sub endTable() ############################################################################## ### end the HTML output ############################################################################## sub endHtml() { -# do not print these links for VGP index - -if ((0 == $vgpIndex) && ($asmHubName ne "viral")) { - printf "

\n\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - printf "\n"; - - printf "\n
Additional hubs with collections of assemblies
Assembly hubs index pages: PrimatesMammalsBirdsFishother vertebrates
Hubs assembly statistics: PrimatesMammalsBirdsFishother vertebrates
Hubs track statistics: PrimatesMammalsBirdsFishother vertebrates
\n

\n"; -} +&commonHtml::otherHubLinks($vgpIndex, $asmHubName); +&commonHtml::htmlFooter($vgpIndex, $asmHubName); -print <<"END" - - - - - -END } # sub endHtml() ############################################################################## ### tableContents() ############################################################################## sub tableContents() { my $rowCount = 0; foreach my $asmId (@orderList) { my ($gcPrefix, $asmAcc, $asmName) = split('_', $asmId, 3); my $accessionId = sprintf("%s_%s", $gcPrefix, $asmAcc); my $accessionDir = substr($asmId, 0 ,3); $accessionDir .= "/" . substr($asmId, 4 ,3); $accessionDir .= "/" . substr($asmId, 7 ,3); $accessionDir .= "/" . substr($asmId, 10 ,3); my $ncbiFtpLink = "ftp://ftp.ncbi.nlm.nih.gov/genomes/all/$accessionDir/$asmId"; my $buildDir = "/hive/data/genomes/asmHubs/refseqBuild/$accessionDir/$asmId"; if ($gcPrefix eq "GCA") { $buildDir = "/hive/data/genomes/asmHubs/genbankBuild/$accessionDir/$asmId"; } my $asmReport="$buildDir/download/${asmId}_assembly_report.txt"; my $trackDb="$buildDir/${asmId}.trackDb.txt"; next if (! -s "$trackDb"); # assembly build not complete my $chromSizes="${buildDir}/${asmId}.chrom.sizes"; my $sciName = "notFound"; my $commonName = "notFound"; my $bioSample = "notFound"; my $bioProject = "notFound"; my $taxId = "notFound"; my $asmDate = "notFound"; my $itemsFound = 0; open (FH, "<$asmReport") or die "can not read $asmReport"; while (my $line = ) { last if ($itemsFound > 5); chomp $line; $line =~ s/ //g;; $line =~ s/\s+$//g;; if ($line =~ m/Date:/) { if ($asmDate =~ m/notFound/) { ++$itemsFound; $line =~ s/.*:\s+//; my @a = split('-', $line); $asmDate = sprintf("%04d-%02d-%02d", $a[0], $a[1], $a[2]); } } elsif ($line =~ m/BioSample:/) { if ($bioSample =~ m/notFound/) { ++$itemsFound; $bioSample = $line; $bioSample =~ s/.*:\s+//; } } elsif ($line =~ m/BioProject:/) { if ($bioProject =~ m/notFound/) { ++$itemsFound; $bioProject = $line; $bioProject =~ s/.*:\s+//; } } elsif ($line =~ m/Organism name:/) { if ($sciName =~ m/notFound/) { ++$itemsFound; $commonName = $line; $sciName = $line; $commonName =~ s/.*\(//; $commonName =~ s/\)//; $commonName = $commonName{$asmId} if (exists($commonName{$asmId})); $sciName =~ s/.*:\s+//; $sciName =~ s/\s+\(.*//; } } elsif ($line =~ m/Taxid:/) { if ($taxId =~ m/notFound/) { ++$itemsFound; $taxId = $line; $taxId =~ s/.*:\s+//; } } } close (FH); my $hubUrl = "https://hgdownload.soe.ucsc.edu/hubs/$accessionDir/$accessionId"; printf "%d\n", ++$rowCount; ### printf "%s\n", $hubUrl, $accessionId, $commonName; printf "%s\n", $accessionId, $commonName; printf " %s\n", $hubUrl, $sciName; printf " %s\n", $gcPrefix, $asmAcc, $asmId; if ( $bioSample ne "notFound" ) { printf " %s\n", $bioSample, $bioSample; } else { printf " n/a\n"; } if ($bioProject eq "notFound") { printf " %s\n", $bioProject; } else { printf " %s\n", $bioProject, $bioProject; } printf " %s\n", $ncbiFtpLink, $asmDate; if ($vgpIndex) { my $sciNameUnderscore = $sciName; $sciNameUnderscore =~ s/ /_/g; $sciNameUnderscore = "Strigops_habroptilus" if ($sciName =~ m/Strigops habroptila/); if (! defined($vgpClass{$asmId})) { printf STDERR "# ERROR: no 'class' defined for VGP assembly %s\n", $asmId; exit 255; } printf " %s\n", $sciNameUnderscore, $vgpClass{$asmId} } printf "\n"; } } # sub tableContents() ############################################################################## ### main() ############################################################################## -open (FH, "<$toolsDir/${commonNameOrder}") or die "can not read ${commonNameOrder}"; +open (FH, "<${orderList}") or die "can not read ${orderList}"; while (my $line = ) { next if ($line =~ m/^#/); chomp $line; my ($asmId, $commonName) = split('\t', $line); push @orderList, $asmId; $commonName{$asmId} = $commonName; ++$assemblyCount; } close (FH); startHtml(); startTable(); tableContents(); endTable(); endHtml();