Mercurial > repos > yusuf > transfer_convert_nextseq

comparison copyNextSeq.pl @ 0:d4ac6e05c96c default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
initial commit
author Yusuf Ali <ali@yusuf.email>
date Wed, 25 Mar 2015 13:43:47 -0600
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:d4ac6e05c96c
1 #!/usr/bin/perl
2 use strict;
3 use warnings;
4 use Getopt::Long;
5 use File::Find;
6 use File::Basename;
7 use vars qw(@fastq_files);
8
9 my $dirname = dirname(__FILE__);
10 my $pythonScript = "$dirname/rgFastQC.py";
11 my $tool_dir = shift @ARGV;
12 my $pythonJars = "$tool_dir/shared/jars/FastQC/fastqc";
13
14 # Site config
15 my $num_threads = 32;
16 my $fastq_sample_size = 400000;
17 my $seq_host = "10.81.192.138";
18 my $seq_username = "nextseq-user";
19 my $seq_dir = "Desktop/Share";
20
21 #get localdir
22 if(not -e "$tool_dir/transfer_convert_nextseq.loc"){
23 system("cat $dirname/tool-data/transfer_convert_nextseq.loc > $tool_dir/transfer_convert_nextseq.loc");
24 }
25 open FILE, "$tool_dir/transfer_convert_nextseq.loc" or die "Could not open configuration file: $!\n";
26 my @keys = split("=",<FILE>);
27 (my $local_dir = $keys[$#keys]) =~s/\s+//g;
28 close FILE;
29
30 # store arguments into variables
31 my $runName;
32 my $sampleSheet;
33 my $user;
34 my $accessFile;
35 my $outDir;
36 my $htmlFile;
37 my $archiveFile;
38
39 GetOptions ("run=s" => \$runName,
40 "samplesheet=s" => \$sampleSheet,
41 "user=s" => \$user,
42 "toolDir=s" => \$accessFile,
43 "out=s" => \$outDir,
44 "html=s" => \$htmlFile,
45 "archive=s" => \$archiveFile);
46
47 if(not defined $runName or not defined $sampleSheet or not defined $user or not defined $accessFile or not defined $outDir or not defined $htmlFile){
48 die "Usage: $0 -run <unique_suffix> -samplesheet <illumina.csv> -user <user\@domain in nextseq_access.conf> -toolDir <galaxy tool conf dir> ",
49 "-out <output dir for FASTQC report> -html <FASTQC report file name> -archive <SAV files.zip>\n";
50 }
51
52 $accessFile = "$accessFile/nextseq_access.conf";
53
54 # create access file if not already there
55 my $command = `touch $accessFile`;
56 open my $handle, '<', "$accessFile";
57 chomp(my @allowed_users = <$handle>);
58
59 $runName = quotemeta($runName);
60
61 my ($out_file, $out_path, $out_ext ) = fileparse( $htmlFile, "\.[^.]*" );
62
63 # check to make sure $user is allowed to run script
64 if (! ($user ~~ @allowed_users) ){
65 die "Please ask the administrator to add $user to $accessFile in order to gain access to this tool\n";
66 }
67
68 # First, sanity check the sample file
69 open(CSV, $sampleSheet)
70 or die "Cannot open $sampleSheet for reading: $!\n";
71 undef $/; # slurp up whole file at once by undefining record separator
72 my @CSV = split /\r?\n/, <CSV>; # allow different endings
73 close(CSV);
74 $/="\n"; # restore normal per-line reading
75 my ($has_header, $has_reads, $has_data);
76 for(@CSV){
77 if(/^\[Header\]/){
78 $has_header = 1;
79 }
80 elsif(/^\[Reads\]/){
81 $has_reads = 1;
82 }
83 elsif(/^\[Data\]/){
84 $has_data = 1;
85 }
86 }
87 if(not defined $has_header){
88 die "Header section is missing in sample sheet, please fix and resubmit this job\n";
89 }
90 if(not defined $has_reads){
91 die "Reads section is missing in sample sheet, please fix and resubmit this job\n";
92 }
93 if(not defined $has_data){
94 die "Data section is missing in sample sheet, please fix and resubmit this job\n";
95 }
96
97 # Expand the catridge ID into the full run name on the remote host, input should look something like "H35VJBGXX"
98 open(SSH, "ssh $seq_username\@$seq_host ls -1 $seq_dir |")
99 or die "Could not run ssh login to $seq_host: $!\n";
100 my @matchOptions;
101 my @mismatchOptions;
102 while(<SSH>){
103 chomp;
104 if(/$runName/o){
105 push @matchOptions, $_;
106 }
107 else{
108 push @mismatchOptions, $_;
109 }
110 }
111 close(SSH);
112 if(not @matchOptions){
113 if(not @mismatchOptions){
114 die "There was no data found on the rempote server at all, please ask the administrator to ",
115 "check this tool's setup (currently checking $seq_username\@$seq_host:$seq_dir)\n";
116 }
117 # Keep only the ones not already uploaded as options
118 @mismatchOptions = grep {not -e "$local_dir/$_"} @mismatchOptions;
119 die "No run folder matching $runName was found at $seq_username\@$seq_host:$seq_dir, please try with another ",
120 "run name. The following would work currently: ", join(", ", @mismatchOptions), "\n";
121 }
122 elsif(@matchOptions > 1){
123 die "Ambiguous run name specification, please revise \"$runName\" to distinguish between existing datasets: ",
124 join(", ", @matchOptions), "\n";
125 }
126 my $expandedRunName = $matchOptions[0]; # unambiguous, so proceed
127
128 # if sample already exits as a folder, die
129 if(-e "$local_dir/$expandedRunName"){
130 # die "Run $expandedRunName already exists on galaxy ($local_dir/$expandedRunName), cannot copy over\n";
131 }
132 # if not, copy to folder
133 else{
134 # 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 $?\n";
135 }
136
137 # Put the sample sheet where it needs to be with the transfered data
138 open(CSV, ">$local_dir/$expandedRunName/SampleSheet.csv")
139 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";
140 print CSV join("\n", @CSV);
141 close(CSV);
142
143 # convert bcl files to fastq
144 #system("cd $local_dir/$expandedRunName; /export/common/programs/bcl2fastq/bin/bcl2fastq -r $num_threads -d $num_threads -p $num_threads -w $num_threads")>>8
145 # and die "BCL to FASTQ conversion had non-zero exit status ($?). The BCL files were transfered, but FASTQ files were not generated.\n";
146
147 # Find the FASTQ files generated
148 find(sub{push @fastq_files, $File::Find::name if /\.fastq.gz$/}, "$local_dir/$expandedRunName");
149
150 # Run FASTQC on sample of data from each lane/barcode
151 # open output file and write html
152 open(OUTFILE, ">$htmlFile")
153 or die "Cannot open $htmlFile for writing: $!\n";
154 print OUTFILE "<html><body><h1>Barcodes</h1>";
155 system("mkdir -p $outDir");
156
157 # generate html plot using python tool
158 $SIG{'PIPE'} = 'IGNORE';
159 my $cwd = dirname(__FILE__);
160 foreach my $file (@fastq_files){
161 my ($barcode, $path, $ext ) = fileparse( $file, "\.fastq\.gz" );
162 my $cmd = "gzip -cd $file | head -n $fastq_sample_size | python $pythonScript -i /dev/stdin "
163 . "-d $outDir/$barcode/. "
164 . "-o fastqc_report.html "
165 . "-n \"FASTQC $barcode\" "
166 . "-f \"FASTQ\" "
167 . "-j \"$barcode$ext\" "
168 . "-e $pythonJars";
169 # Assumes the bash shell is being used
170 open(CMD, "trap '' SIGPIPE; $cmd 2| grep -v \"Broken pipe\" |")
171 or die "Cannot run FASTQC: $!\n";
172 while(<CMD>){
173 # Can safely ignore blank lines and SIGPIPE warnings
174 next if /^\s*$/ or /Broken pipe/;
175 print STDERR $_; # forward any other errors
176 }
177 close(CMD);
178 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");
179 system("unzip -o -d $outDir/$barcode -qq -j $outDir/$barcode/$barcode\_fastqc.zip $barcode\_fastqc/Icons/*.png");
180 # append to html file
181 print OUTFILE "<div><a href='$barcode/fastqc_report.html'>$barcode</a></div>";
182 }
183
184
185 print OUTFILE "</body></html>";
186 close(OUTFILE);
187 system("cp $htmlFile $outDir/index.html");
188 system("cd $local_dir/$expandedRunName; rm $archiveFile; zip -r $archiveFile RunInfo.xml RunParameters.xml InterOp -q");