diff copyNextSeq.pl @ 0:d4ac6e05c96c default tip

initial commit

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/copyNextSeq.pl	Wed Mar 25 13:43:47 2015 -0600
@@ -0,0 +1,188 @@
+#!/usr/bin/perl
+use strict;
+use warnings;
+use Getopt::Long;
+use File::Find;
+use File::Basename;
+use vars qw(@fastq_files);
+
+my $dirname = dirname(__FILE__);
+my $pythonScript = "$dirname/rgFastQC.py";
+my $tool_dir = shift @ARGV;
+my $pythonJars = "$tool_dir/shared/jars/FastQC/fastqc";
+
+# Site config
+my $num_threads = 32;
+my $fastq_sample_size = 400000;
+my $seq_host = "10.81.192.138";
+my $seq_username = "nextseq-user";
+my $seq_dir = "Desktop/Share";
+
+#get localdir
+if(not -e "$tool_dir/transfer_convert_nextseq.loc"){
+  system("cat $dirname/tool-data/transfer_convert_nextseq.loc > $tool_dir/transfer_convert_nextseq.loc");
+}
+open FILE, "$tool_dir/transfer_convert_nextseq.loc" or die "Could not open configuration file: $!\n";
+my @keys = split("=",<FILE>);
+(my $local_dir = $keys[$#keys]) =~s/\s+//g;
+close FILE;
+
+# store arguments into variables
+my $runName;
+my $sampleSheet;
+my $user;
+my $accessFile;
+my $outDir;
+my $htmlFile;
+my $archiveFile;
+
+GetOptions ("run=s" => \$runName,
+            "samplesheet=s" => \$sampleSheet,
+	    "user=s" => \$user,
+	    "toolDir=s" => \$accessFile,
+	    "out=s" => \$outDir,
+	    "html=s" => \$htmlFile,
+            "archive=s" => \$archiveFile);
+
+if(not defined $runName or not defined $sampleSheet or not defined $user or not defined $accessFile or not defined $outDir or not defined $htmlFile){
+  die "Usage: $0 -run <unique_suffix> -samplesheet <illumina.csv> -user <user\@domain in nextseq_access.conf> -toolDir <galaxy tool conf dir> ",
+      "-out <output dir for FASTQC report> -html <FASTQC report file name> -archive <SAV files.zip>\n";
+}
+
+$accessFile = "$accessFile/nextseq_access.conf";
+
+# create access file if not already there
+my $command = `touch $accessFile`;
+open my $handle, '<', "$accessFile";
+chomp(my @allowed_users = <$handle>);
+
+$runName = quotemeta($runName);
+
+my ($out_file, $out_path, $out_ext ) = fileparse( $htmlFile, "\.[^.]*" );
+
+# check to make sure $user is allowed to run script
+if (! ($user ~~ @allowed_users) ){
+	die "Please ask the administrator to add $user to $accessFile in order to gain access to this tool\n";
+}
+
+# First, sanity check the sample file
+open(CSV, $sampleSheet)
+  or die "Cannot open $sampleSheet for reading: $!\n";
+undef $/; # slurp up whole file at once by undefining record separator
+my @CSV = split /\r?\n/, <CSV>; # allow different endings
+close(CSV);
+$/="\n"; # restore normal per-line reading
+my ($has_header, $has_reads, $has_data);
+for(@CSV){
+  if(/^\[Header\]/){
+    $has_header = 1;
+  }
+  elsif(/^\[Reads\]/){
+    $has_reads = 1;
+  }
+  elsif(/^\[Data\]/){
+    $has_data = 1;
+  }
+}
+if(not defined $has_header){
+	die "Header section is missing in sample sheet, please fix and resubmit this job\n";
+}
+if(not defined $has_reads){
+	die "Reads section is missing in sample sheet, please fix and resubmit this job\n";
+}
+if(not defined $has_data){
+	die "Data section is missing in sample sheet, please fix and resubmit this job\n";
+}
+
+# Expand the catridge ID into the full run name on the remote host, input should look something like "H35VJBGXX"
+open(SSH, "ssh $seq_username\@$seq_host ls -1 $seq_dir |")
+  or die "Could not run ssh login to $seq_host: $!\n";
+my @matchOptions;
+my @mismatchOptions;
+while(<SSH>){
+	chomp;
+	if(/$runName/o){
+		push @matchOptions, $_;	
+	}
+	else{
+		push @mismatchOptions, $_;
+	}
+}
+close(SSH);
+if(not @matchOptions){
+	if(not @mismatchOptions){
+		die "There was no data found on the rempote server at all, please ask the administrator to ",
+		    "check this tool's setup (currently checking $seq_username\@$seq_host:$seq_dir)\n";
+	}
+	# Keep only the ones not already uploaded as options
+	@mismatchOptions = grep {not -e "$local_dir/$_"} @mismatchOptions;
+	die "No run folder matching $runName was found at $seq_username\@$seq_host:$seq_dir, please try with another ",
+	    "run name. The following would work currently: ", join(", ", @mismatchOptions), "\n";  
+}
+elsif(@matchOptions > 1){
+	die "Ambiguous run name specification, please revise \"$runName\" to distinguish between existing datasets: ",
+	    join(", ", @matchOptions), "\n"; 
+}
+my $expandedRunName = $matchOptions[0]; # unambiguous, so proceed
+
+# if sample already exits as a folder, die
+if(-e "$local_dir/$expandedRunName"){
+#	die "Run $expandedRunName already exists on galaxy ($local_dir/$expandedRunName), cannot copy over\n";
+}
+# if not, copy to folder
+else{
+#	system("scp -r $seq_username\@$seq_host\:$seq_dir/$expandedRunName $local_dir") >> 8 and die "Failed to copy from $seq_host to galaxy: scp exit status $?\n";
+}
+
+# Put the sample sheet where it needs to be with the transfered data
+open(CSV, ">$local_dir/$expandedRunName/SampleSheet.csv")
+  or die "Cannot open $local_dir/$expandedRunName/SampleSheet.csv for writing: $!\nThe data files have been transfered, but no BCL to FASTQ conversion has taken place.\n";
+print CSV join("\n", @CSV);
+close(CSV);
+
+# convert bcl files to fastq
+#system("cd $local_dir/$expandedRunName; /export/common/programs/bcl2fastq/bin/bcl2fastq -r $num_threads -d $num_threads -p $num_threads -w $num_threads")>>8
+#  and die "BCL to FASTQ conversion had non-zero exit status ($?). The BCL files were transfered, but FASTQ files were not generated.\n";
+
+# Find the FASTQ files generated
+find(sub{push @fastq_files, $File::Find::name if /\.fastq.gz$/}, "$local_dir/$expandedRunName");
+
+# Run FASTQC on sample of data from each lane/barcode
+# open output file and write html
+open(OUTFILE, ">$htmlFile")
+  or die "Cannot open $htmlFile for writing: $!\n";
+print OUTFILE "<html><body><h1>Barcodes</h1>";
+system("mkdir -p $outDir");
+
+# generate html plot using python tool
+$SIG{'PIPE'} = 'IGNORE'; 
+my $cwd = dirname(__FILE__);
+foreach my $file (@fastq_files){
+	my ($barcode, $path, $ext ) = fileparse( $file, "\.fastq\.gz" );
+	my $cmd = "gzip -cd $file | head -n $fastq_sample_size | python $pythonScript -i /dev/stdin "
+. "-d $outDir/$barcode/. "
+. "-o fastqc_report.html "
+. "-n \"FASTQC $barcode\" "
+. "-f \"FASTQ\" "
+. "-j \"$barcode$ext\" "
+. "-e $pythonJars"; 
+	# Assumes the bash shell is being used
+	open(CMD, "trap '' SIGPIPE; $cmd 2| grep -v \"Broken pipe\" |")
+          or die "Cannot run FASTQC: $!\n";
+        while(<CMD>){
+          # Can safely ignore blank lines and SIGPIPE warnings
+          next if /^\s*$/ or /Broken pipe/; 
+          print STDERR $_; # forward any other errors
+        }
+        close(CMD);
+	system("perl -i.bak -pe \"s/>FastQC Report</>FastQC Report<div><a href='..\\/index.html'>Back to Table of Contents<\\\/a><\\\/div></;s/Images|Icons/./\" $outDir/$barcode/fastqc_report.html");
+        system("unzip -o -d $outDir/$barcode -qq -j $outDir/$barcode/$barcode\_fastqc.zip $barcode\_fastqc/Icons/*.png");
+	# append to html file
+	print OUTFILE "<div><a href='$barcode/fastqc_report.html'>$barcode</a></div>";
+}
+
+
+print OUTFILE "</body></html>";
+close(OUTFILE);
+system("cp $htmlFile $outDir/index.html");
+system("cd $local_dir/$expandedRunName; rm $archiveFile; zip -r $archiveFile RunInfo.xml RunParameters.xml InterOp -q");