951ae58cfc4a8b6b7dec665cbabb524abb4fc67a hiram Fri May 1 13:09:53 2020 -0700 add class column in the table for VGP genomes refs #23734 diff --git src/hg/makeDb/doc/asmHubs/mkHubIndex.pl src/hg/makeDb/doc/asmHubs/mkHubIndex.pl index 724f50d..54600b2 100755 --- src/hg/makeDb/doc/asmHubs/mkHubIndex.pl +++ src/hg/makeDb/doc/asmHubs/mkHubIndex.pl @@ -4,30 +4,42 @@ use warnings; my $argc = scalar(@ARGV); if ($argc != 3) { printf STDERR "mkAsmStats Name asmName\n"; printf STDERR "e.g.: mkHubIndex Primates primates GCF_000001405.39_GRCh38.p13\n"; exit 255; } my $Name = shift; my $asmHubName = shift; my $defaultAssembly = shift; my $home = $ENV{'HOME'}; my $toolsDir = "$home/kent/src/hg/makeDb/doc/asmHubs"; my $commonNameOrder = "$asmHubName.commonName.asmId.orderList.tsv"; +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"; + open (FH, "<$vgpClass") or die "can not read $vgpClass"; + while (my $line = <FH>) { + my ($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 $assemblyCount = 0; my %betterName; # key is asmId, value is a common name better than found # in assembly_report file ############################################################################## # from Perl Cookbook Recipe 2.17, print out large numbers with comma delimiters: ############################################################################## sub commify($) { my $text = reverse $_[0]; $text =~ s/(\d\d\d)(?=\d)(?!\d*\.)/$1,/g; return scalar reverse $text } @@ -35,31 +47,31 @@ ############################################################################## ### start the HTML output ############################################################################## sub startHtml() { my $timeStamp = `date "+%F"`; chomp $timeStamp; # <html xmlns="http://www.w3.org/1999/xhtml"> my $subSetMessage = "subset of $asmHubName only"; if ($asmHubName eq "vertebrate") { $subSetMessage = "subset of other ${asmHubName}s only"; } -if ($Name =~ m/vgp/i) { +if ($vgpIndex) { print <<"END" <!DOCTYPE HTML 4.01 Transitional> <!--#set var="TITLE" value="VGP - Vertebrate Genomes Project assembly hub" --> <!--#set var="ROOT" value="../.." --> <!--#include virtual="\$ROOT/inc/gbPageStartHardcoded.html" --> <h1>VGP - Vertebrate Genomes Project assembly hub</h1> <p> <a href='https://vertebrategenomesproject.org/' target=_blank> <img src='VGPlogo.png' width=280 alt='VGP logo'></a></p> <p> This assembly hub contains assemblies released by the <a href='https://vertebrategenomesproject.org/' target=_blank> Vertebrate Genomes Project.</a> @@ -110,59 +122,64 @@ <li>Then select the <strong>My Hubs</strong> tab and enter this URL into the textbox: <br><code>https://hgdownload.soe.ucsc.edu/hubs/$asmHubName/hub.txt</code></li> <li> Once you have added the URL to the entry form, press the <em><strong>Add Hub</strong></em> button to add the hub.</li> </ol> </li> </ol> </p> <p> After adding the hub, you will be redirected to the gateway page. The genome assemblies can be selected from the <em>${Name} Hub Assembly</em> dropdown menu. Instead of adding all the assemblies in one collected group, use the individual -<em>link to genome browser</em> in the table below. +<em>view in browser</em> in the table below. </p> <h3>See also: <a href='asmStats.html'>assembly statistics</a>, <a href='trackData.html'>track statistics</a> <== additional information for these assemblies.</h3><br> <h3>Data resource links</h3> <p> NOTE: <em>Click on the column headers to sort the table by that column</em><br> -The <em>common name/link to genome browser</em> will attach only that single assembly to +The <em>common name/view in browser</em> will attach only that single assembly to the genome browser.<br> The <em>scientific name/and data download</em> link provides access to the files for that one assembly hub.<br> +The <em>class/VGP link</em> provides access to the VGP GenomeArk page for that genome<br> The other links provide access to NCBI resources for these assemblies. </p> END } # sub startHtml() ############################################################################## ### start the table output ############################################################################## sub startTable() { -print <<"END" +print ' <table class="sortable" border="1"> <thead><tr><th>count</th> - <th>common name<br>link to genome browser</th> + <th>common name and<br>view in browser</th> <th>scientific name<br>and data download</th> <th>NCBI assembly</th> <th>bioSample</th><th>bioProject</th> <th>assembly date,<br>source link</th> -</tr></thead><tbody> -END +'; + +if ($vgpIndex) { + printf "<th>class<br>VGP link</th>\n"; +} +print "</tr></thead><tbody>\n"; } # sub startTable() ############################################################################## ### end the table output ############################################################################## sub endTable() { print <<"END" </tbody> </table> END } # sub endTable() ############################################################################## @@ -282,30 +299,37 @@ } close (FH); my $hubUrl = "https://hgdownload.soe.ucsc.edu/hubs/$accessionDir/$accessionId"; printf "<tr><td align=right>%d</td>\n", ++$rowCount; ### printf "<td align=center><a href='https://genome.ucsc.edu/cgi-bin/hgGateway?hubUrl=%s/hub.txt&genome=%s&position=lastDbPos' target=_blank>%s</a></td>\n", $hubUrl, $accessionId, $commonName; printf "<td align=center><a href='https://genome.ucsc.edu/h/%s' target=_blank>%s</a></td>\n", $accessionId, $commonName; printf " <td align=center><a href='%s/' target=_blank>%s</a></td>\n", $hubUrl, $sciName; printf " <td align=left><a href='https://www.ncbi.nlm.nih.gov/assembly/%s_%s/' target=_blank>%s</a></td>\n", $gcPrefix, $asmAcc, $asmId; if ( $bioSample ne "notFound" ) { printf " <td align=left><a href='https://www.ncbi.nlm.nih.gov/biosample/?term=%s' target=_blank>%s</a></td>\n", $bioSample, $bioSample; } else { printf " <td align=left>n/a</td>\n"; } printf " <td align=left><a href='https://www.ncbi.nlm.nih.gov/bioproject/?term=%s' target=_blank>%s</a></td>\n", $bioProject, $bioProject; printf " <td align=center><a href='%s' target=_blank>%s</a></td>\n", $ncbiFtpLink, $asmDate; + if ($vgpIndex) { + my $sciNameUnderscore = $sciName; + $sciNameUnderscore =~ s/ /_/g; + $sciNameUnderscore = "Strigops_habroptilus" if ($sciName =~ m/Strigops habroptila/); + + printf " <td align=center><a href='https://vgp.github.io/genomeark/%s/' target=_blank>%s</a></td>\n", $sciNameUnderscore, $vgpClass{$asmId} + } printf "</tr>\n"; } } # sub tableContents() ############################################################################## ### main() ############################################################################## 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; $betterName{$asmId} = $commonName;