d3219932529dacd180646575f4892e8a32ba2601
hiram
Mon Feb 24 10:16:46 2020 -0800
adding trackData page construction refs #23891
diff --git src/hg/makeDb/doc/asmHubs/trackData.pl src/hg/makeDb/doc/asmHubs/trackData.pl
new file mode 100755
index 0000000..074214a
--- /dev/null
+++ src/hg/makeDb/doc/asmHubs/trackData.pl
@@ -0,0 +1,360 @@
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+use File::stat;
+
+my $argc = scalar(@ARGV);
+if ($argc != 2) {
+ printf STDERR "usage: trackData.pl Name asmHubName > trackData.html\n";
+ printf STDERR "e.g.: trackData.pl Mammals mammals > trackData.html\n";
+ exit 255;
+}
+my $Name = shift;
+my $asmHubName = shift;
+
+my $home = $ENV{'HOME'};
+my $toolsDir = "$home/kent/src/hg/makeDb/doc/asmHubs";
+
+my $commonNameList = "$asmHubName.asmId.commonName.tsv";
+my $commonNameOrder = "$asmHubName.commonName.asmId.orderList.tsv";
+my @orderList; # asmId of the assemblies in order from the *.list files
+# the order to read the different .list files:
+my %betterName; # key is asmId, value is better common name than found in
+ # assembly_report
+
+my $assemblyTotal = 0; # complete list of assemblies in this group
+my $asmCount = 0; # count of assemblies completed and in the table
+my $overallNucleotides = 0;
+my $overallSeqCount = 0;
+my $overallGapSize = 0;
+my $overallGapCount = 0;
+
+##############################################################################
+# from Perl Cookbook Recipe 2.17, print out large numbers with comma delimiters:
+##############################################################################
+sub commify($) {
+ my $text = reverse $_[0];
+ $text =~ s/(\d\d\d)(?=\d)(?!\d*\.)/$1,/g;
+ return scalar reverse $text
+}
+
+sub oneTrackData($$) {
+ my ($file, $genomeSize) = @_;
+# printf STDERR "# %s\n", $file;
+ my $itemCount = 0;
+ my $percentCover = 0;
+ if (! -s "${file}") {
+ return("n/a", "n/a");
+ }
+ if ($file =~ m/.bw$/) {
+ my $bigWigInfo = `bigWigInfo "$file" | egrep "basesCovered:|mean:" | awk '{print \$NF}' | xargs echo | sed -e 's/,//g;'`;
+ chomp $bigWigInfo;
+ my ($bases, $mean) = split('\s+', $bigWigInfo);
+ $percentCover = sprintf("%.2f %%", 100.0 * $bases / $genomeSize);
+ $itemCount = sprintf ("%.2f", $mean);
+# printf STDERR "# bigWigInfo %s %s %s\n", $itemCount, $percentCover, $file;
+ } else {
+ my $bigBedInfo = `bigBedInfo "$file" | egrep "itemCount:|basesCovered:" | awk '{print \$NF}' | xargs echo | sed -e 's/,//g;'`;
+ chomp $bigBedInfo;
+ my ($items, $bases) = split('\s', $bigBedInfo);
+ $itemCount = commify($items);
+ $percentCover = sprintf("%.2f %%", 100.0 * $bases / $genomeSize);
+# printf STDERR "# bigBedInfo %s %s %s\n", $itemCount, $percentCover, $file;
+ }
+ return ($itemCount, $percentCover);
+}
+
+##############################################################################
+### 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 ${asmHubName}s only";
+}
+
+print <<"END"
+
+
+
+
+
+
+$Name Genomes assembly hubs
+
+Assemblies from NCBI/Genbank/Refseq sources, $subSetMessage.
+
+
+
+
+Data resource links
+NOTE: Click on the column headers to sort the table by that column
+The link to genome browser will attach only that single assembly to
+the genome browser.
+The numbers are: item count (percent coverage)
+Except for the gc5Base column which is: overall GC % average (percent coverage)
+END
+}
+
+##############################################################################
+### start the table output
+##############################################################################
+sub startTable() {
+print <<"END"
+
+| count |
+ common name
link to genome browser |
+ gc5 base |
+ gap |
+ assembly |
+ rmsk |
+ TRF
simpleRepeat |
+ windowMasker |
+ gapOverlap |
+ tandemDups |
+ cpg
unmasked |
+ cpg
island |
+ ncbiGene |
+ ncbiRefSeq |
+ xenoRefGene |
+ augustus |
+
|---|
+END
+}
+
+##############################################################################
+### end the table output
+##############################################################################
+sub endTable() {
+
+my $commaNuc = commify($overallNucleotides);
+my $commaSeqCount = commify($overallSeqCount);
+my $commaGapSize = commify($overallGapSize);
+my $commaGapCount = commify($overallGapCount);
+
+my $percentDone = 100.0 * $asmCount / $assemblyTotal;
+
+if ($assemblyTotal > 1) {
+ print <<"END"
+
+
+| TOTALS: | total assembly count ${assemblyTotal} |
+
|---|
+
+END
+} else {
+ print <<"END"
+
+
+
+END
+}
+} # sub endTable()
+
+##############################################################################
+### end the HTML output
+##############################################################################
+sub endHtml() {
+
+if ($asmHubName ne "viral") {
+ printf "\nOther assembly hubs available:
\n
\n";
+
+ printf "| Primates | \n"
+ if ($asmHubName ne "primates");
+ printf "Mammals | \n"
+ if ($asmHubName ne "mammals");
+ printf "Birds | \n"
+ if ($asmHubName ne "birds");
+ printf "Fish | \n"
+ if ($asmHubName ne "fish");
+ printf "other vertebrates | \n"
+ if ($asmHubName ne "vertebrate");
+
+ printf "
\n
\n\n";
+}
+
+print <<"END"
+
+
+
+
+