\n"; printf "

\n"; - -printf "

\n"; -printf " show legend for column definitions\n"; -printf "

\n"; -printf "

Click on the column headers to sort the table by that column
The common name column will either open a genome browser on that assembly when it exists, or it will submit a request to have this assembly added to the GenArk collection when it has not yet been built.
Use the select columns of information to display menu to show more or less information in the tables.
Use the select assembly type display menu to refine the types of assemblies to show in the tables.
Use the show/hide species clade sets menu to refine the species clades to display.
Use the hide all selection to hide everything, or restore all items back when selections have been partially hidden.
The scientific name link opens google image search for that species. A name followed by a number in (parens) indicates number of assemblies available for this species when there are too many to include here.
IUCN status column is the status from the IUCN (2021) Red List of Threatened species, Version 2021-3, accessed on 17 Dec 2021 CR - Critical / EN - Endangered / VU - Vulnerable The link goes to the IUCN web site for more information for that species.
The NCBI taxID links to the NCBI taxonomy database.

\n"; -printf "

\n"; - - printf "

\n"; printf " choose clades to view/hide\n"; printf "

\n"; printf " \n"; printf "

\n"; @@ -805,42 +786,42 @@ 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 " common name
display in browser
or request assembly\n"; -printf " scientific name (count)
google image search\n"; -printf " NCBI assembly\n"; -printf " assembly
size\n"; -printf " sequence
count\n"; -printf " scaffold N50
size (L50)\n"; -printf " contig N50
size (L50)\n"; -printf " I
U
C
N\n"; -printf " NCBI taxID\n"; -printf " assembly
date\n"; -printf " submitter of assembly
link to NCBI assembly\n"; -printf " clade\n"; +printf " \n"; +printf " \n"; +printf " \n"; +printf " \n"; +printf " \n"; +printf " \n"; +printf " \n"; +printf " \n", $statusColors{"CR"}, $statusColors{"EN"}, $statusColors{"VU"}; +printf " \n"; +printf " \n"; +printf " \n"; +printf " \n"; printf "\n"; printf "\n"; my %equivalentNamesUsed; # key is NCBI sciName, value is IUCN sciName my $pageSectionCount = 0; my %checkDupAsmId; # key is asmId, value is count of times seen my %cladeSciNameCounts; # key is clade, value is number of different # scientific names my %gcfGcaCounts; # key is GCF or GCA, value is count of each my $asmCountInTable = 0; # counting the rows output my %statusCounts; # key is status: CR EN VU, value is count my $totalAssemblySize = 0; # sum of all assembly sizes @@ -961,31 +942,31 @@ ($n50ScaffoldSize, $n50ScaffoldCount) = readN50($scaffoldN50); } } else { printf STDERR "# chromInfo missing: %s\n", $asmId; printf STDERR "# %s\n", $chromInfo; } die "ERROR duplicate newMetaInfo{$asmId}" if (defined($newMetaInfo{$asmId})); die "ERROR duplicate metaInfo{$asmId} for newMetaInfo{$asmId}" if (defined($metaInfo{$asmId})); $newMetaInfo{$asmId} = join("\t", ($asmSize, $asmContigCount, $n50Size, $n50Count, $n50ContigSize, $n50ContigCount, $n50ScaffoldSize, $n50ScaffoldCount) ); } # else if (defined($metaInfo{$asmId})) # if asmSize is below the minimum, don't use it if ($asmSize < $minimalGenomeSize{$clade}) { printf STDERR "# %s underSized 2 %d %s %s < %s\n", $clade, ++$underSized, $asmId, commify($asmSize), commify($minimalGenomeSize{$clade}); printf STDERR "# ACK would be genArk assembly %s\n", $asmId if (defined($genArkAsm{$asmId})); printf STDERR "# ACK would be UCSC RR %s\n", $asmId if (defined($rrGcaGcfList{$asmId})); -### XXX next; + next; } my $iucnStatus = " "; my $iucnLink = ""; if (defined($sciNames{$asmId})) { my $iucnSciName = $sciNames{$asmId}; $iucnSciName = $ncbiToIucnNames{$sciNames{$asmId}} if (defined($ncbiToIucnNames{$sciNames{$asmId}})); $iucnLink = "https://www.iucnredlist.org/species/$iucnLink{$iucnSciName}" if (defined($iucnLink{$iucnSciName})); if ($iucnSciName ne $sciNames{$asmId}) { $equivalentNamesUsed{$sciNames{$asmId}} = $iucnSciName; } if (defined($iucnSciNames{$iucnSciName})) { $iucnStatus = $iucnSciNames{$iucnSciName}; } } ++$asmCountInTable; @@ -1139,34 +1120,36 @@ printf PC "\t%s", "n/a"; # output to clade.tableData.txt } ############# tenth column, assembly date ################ if (defined($asmDate{$asmId})) { printf "%s", $asmDate{$asmId}; printf PC "\t%s", $asmDate{$asmId}; # output to clade.tableData.txt } else { printf "n/a"; printf PC "\t%s", "n/a"; # output to clade.tableData.txt } ############# eleventh column, submitter ################ $asmUrl = "https://www.ncbi.nlm.nih.gov/assembly/$accessionId"; if (defined($asmSubmitter{$asmId})) { - printf "%s", $asmUrl, $asmSubmitter{$asmId}; + my $submitterSortKey = lc($asmSubmitter{$asmId}); + $submitterSortKey =~ s/ //g; + printf "%s", substr($submitterSortKey,0,20), $asmSubmitter{$asmId}; printf PC "\t%s", $asmSubmitter{$asmId}; # output to clade.tableData.txt } else { - printf "n/a"; + printf "n/a"; printf PC "\t%s", "n/a"; # output to clade.tableData.txt } ############# twelveth column, clade ################ printf "%s\n", $clade; printf PC "\t%s", $clade; printf PC "\n"; # finished a line output to clade.tableData.txt printf "\n"; } # foreach my $asmId (@$cPtr) close (PC); # finished with clade.tableData.txt output } # foreach my $clade (@clades) ########################################################################## ## single table is finished, output the end of tbody and the tfoot row