Merge pull request #57 from ICGC-TCGA-PanCancer/dev

Roll 1.3.0
ICGC-TCGA-PanCancer · Feb 5, 2015 · b00b28f · b00b28f
2 parents 7cb9d0a + 8078ad6
commit b00b28f
Show file tree

Hide file tree

Showing 10 changed files with 172 additions and 59 deletions.
diff --git a/MYMETA.json b/MYMETA.json
@@ -4,7 +4,7 @@
       "unknown"
    ],
    "dynamic_config" : 0,
-   "generated_by" : "ExtUtils::MakeMaker version 6.68, CPAN::Meta::Converter version 2.142690",
+   "generated_by" : "ExtUtils::MakeMaker version 6.68, CPAN::Meta::Converter version 2.131560",
    "license" : [
       "gpl_2"
    ],

diff --git a/MYMETA.yml b/MYMETA.yml
@@ -3,38 +3,38 @@ abstract: unknown
 author:
   - unknown
 build_requires:
-  ExtUtils::MakeMaker: '0'
+  ExtUtils::MakeMaker: 0
 configure_requires:
-  ExtUtils::MakeMaker: '0'
+  ExtUtils::MakeMaker: 0
 dynamic_config: 0
-generated_by: 'ExtUtils::MakeMaker version 6.68, CPAN::Meta::Converter version 2.142690'
+generated_by: 'ExtUtils::MakeMaker version 6.68, CPAN::Meta::Converter version 2.131560'
 license: gpl
 meta-spec:
   url: http://module-build.sourceforge.net/META-spec-v1.4.html
-  version: '1.4'
+  version: 1.4
 name: PCAP
 no_index:
   directory:
     - t
     - inc
 requires:
-  Bio::DB::Sam: '1.39'
-  Bio::Root::Version: '1.006923'
-  Capture::Tiny: '0.24'
-  Const::Fast: '0.014'
-  Data::UUID: '1.219'
-  Devel::Cover: '1.09'
-  File::Which: '0.05'
-  GD: '2.52'
-  IPC::System::Simple: '1.25'
-  List::Util: '1.38'
-  Math::Gradient: '0.04'
-  Module::Build: '0.42'
-  Pod::Coverage: '0.23'
-  Proc::ProcessTable: '0.5'
-  Sub::Exporter::Progressive: '0.001011'
-  Term::UI: '0.42'
-  Test::Fatal: '0.013'
-  Try::Tiny: '0.19'
-  XML::Simple: '2.2'
+  Bio::DB::Sam: 1.39
+  Bio::Root::Version: 1.006923
+  Capture::Tiny: 0.24
+  Const::Fast: 0.014
+  Data::UUID: 1.219
+  Devel::Cover: 1.09
+  File::Which: 0.05
+  GD: 2.52
+  IPC::System::Simple: 1.25
+  List::Util: 1.38
+  Math::Gradient: 0.04
+  Module::Build: 0.42
+  Pod::Coverage: 0.23
+  Proc::ProcessTable: 0.5
+  Sub::Exporter::Progressive: 0.001011
+  Term::UI: 0.42
+  Test::Fatal: 0.013
+  Try::Tiny: 0.19
+  XML::Simple: 2.2
 version: v1.2.3
diff --git a/bin/bam_stats.pl b/bin/bam_stats.pl
@@ -34,6 +34,14 @@ BEGIN
 use File::Spec;
 use Try::Tiny;
 use Pod::Usage qw(pod2usage);
+use Config; # so we can see if threads are enabled
+use Data::Dumper;
+
+BEGIN {
+  if($Config{useithreads}) {
+    require threads;
+  }
+};
 
 use PCAP::Bam::Stats;
 
@@ -52,14 +60,39 @@ BEGIN
   }
 
   try{
-    my $stats = new PCAP::Bam::Stats(-path => $opts->{'input'}, -qscoring => defined $plots_dir);
+    my $stats;
+    if($opts->{'threads'} > 1 && $Config{useithreads}) {
+      for my $thread(0..($opts->{'threads'}-1)) {
+         my ($thr) = threads->create(\&stat_thread, $opts, $thread);
+      }
+      sleep 2 while(threads->list(threads::running) > 0);
+      my @bas_objs;
+      for my $thr(threads->list(threads::joinable)) {
+        push @bas_objs, $thr->join;
+        if(my $err = $thr->error) { die "Thread error: $err\n"; }
+      }
+      # something to merge the stats into a single element of the object
+      $stats = PCAP::Bam::Stats->new(-path => $opts->{'input'},
+                                     -no_proc => 1);
+      $stats->merge_json_stats(\@bas_objs);
+    }
+    else {
+      $stats = PCAP::Bam::Stats->new(-path => $opts->{'input'},
+                                     -qscoring => defined $plots_dir);
+      $stats->fqplots($plots_dir) if($plots_dir);
+    }
     $stats->bas($output);
-    $stats->fqplots($plots_dir) if($plots_dir);
-  }catch{
+  } catch{
     die 'Reading: |'.$opts->{'input'}."| Writing to: |$out_location| Error: $_";
-  }finally{
+  } finally{
     close $output or die "Unable to close |".$opts->{'output'}."|: $!" if($opts->{'output'});
-  }
+  };
+}
+
+sub stat_thread {
+  my ($opts, $thread) = @_;
+  my $stats = PCAP::Bam::Stats->new(-path => $opts->{'input'}, -mod => $opts->{'threads'}, -rem => $thread);
+  return $stats->json_stats;
 }
 
 sub setup{
@@ -70,6 +103,7 @@ sub setup{
               'v|version' => \$opts{'v'},
               'i|input=s' => \$opts{'input'},
               'o|output=s' => \$opts{'output'},
+              't|threads=i' => \$opts{'threads'},
               'p|plots=s' => \$opts{'plots'},
               '<>' => sub{push(@random_args,shift(@_));}
   ) or pod2usage(2);
@@ -88,6 +122,15 @@ sub setup{
   pod2usage(-message  => "\nERROR: i|input must be defined.\n", -verbose => 1,  -output => \*STDERR) unless($opts{'input'});
   pod2usage(-message  => "\nERROR: i|input |".$opts{'input'}."| must be a valid file.\n", -verbose => 1,  -output => \*STDERR) unless(-f $opts{'input'});
 
+  # technically can be done but not a priority
+  pod2usage(-message  => "\nERROR: '-p' and '-t' cannot be used together\n", -verbose => 1,  -output => \*STDERR) if(defined $opts{'plots'} && defined $opts{'threads'});
+
+  $opts{'threads'} = 1 unless(defined $opts{'threads'});
+  if(!$Config{useithreads} && $opts{'threads'} > 1) {
+    warn qq{\nNOTE: Perl not compiled with threads enabled, running un-threaded\n};
+    $opts{'threads'} = 1;
+  }
+
   return \%opts;
 }
 
@@ -114,7 +157,7 @@ =head1 SYNOPSIS
     -version  -v   Prints the version number.
 
     bam_stats.pl -i my.bam -o my.bam.bas
-    cat my.bam | bam_stats.pl > my.bam.bas
+    bam_stats.pl -i my.bam > my.bam.bas
 
 =head1 OPTIONS
 

diff --git a/bin/build_biobambam_relocatable.sh b/bin/build_biobambam_relocatable.sh
@@ -1,8 +1,8 @@
 #! /bin/bash
-LIBMAUSVERSION=0.0.167-release-20140917101853
-BIOBAMBAMVERSION=0.0.171-release-20140917102804
+LIBMAUSVERSION=0.0.191-release-20150129231009
+BIOBAMBAMVERSION=0.0.185-release-20150116094537
 SNAPPYVERSION=1.1.2
-IOLIBVERSION=1.13.7
+IOLIBVERSION=1.13.9
 CHRPATHVERSION=0.16
 BUILDDIR=${PWD}
 INSTALLDIR=${BUILDDIR}/install-dir

diff --git a/bin/diff_bams.pl b/bin/diff_bams.pl
@@ -35,7 +35,7 @@ BEGIN
 
 use Bio::DB::Sam;
 
-my ($file_a, $file_b, $skip_z) = setup();
+my ($file_a, $file_b, $skip_z, $count_flag_diff) = setup();
 
 my $bam_a = Bio::DB::Bam->open($file_a);
 my $bam_b = Bio::DB::Bam->open($file_b);
@@ -59,6 +59,7 @@ BEGIN
 print "Reference sequence order passed\n";
 
 my ($align_a, $align_b, $count);
+my $flag_diffs = 0;
 $|++;
 while(1) {
   $count++;
@@ -74,19 +75,34 @@ BEGIN
   die "Files have different number of records\n" if((!defined $align_a && defined $align_b) || (defined $align_a && !defined $align_b));
 
   die sprintf "Files differ at record $count (qname) a=%s b=%s\n", $align_a->qname, $align_b->qname if($align_a->qname ne $align_b->qname);
-  die sprintf "Files differ at record $count (flags) a=%s b=%s\n", $align_a->qname, $align_b->qname if($align_a->flag ne $align_b->flag);
+  if($align_a->flag ne $align_b->flag) {
+    if($count_flag_diff) {
+      $flag_diffs++;
+      print $align_a->pos."\n";
+    }
+    else {
+      die sprintf "Files differ at record $count (flags) a=%s b=%s\n", $align_a->qname, $align_b->qname
+    }
+  }
+
+  if($count % 100_000 == 0) {
+    my $message = "Matching records: $count";
+    $message .= "\t(flag mismatch: $flag_diffs)"if($count_flag_diff);
+    print $message."\r";
+  }
 
-  print "Matching records: $count\r" if($count % 100_000 == 0);
 }
 print "Matching records: $count\n";
-print "Files match\n";
+print "Flag mismatches: $flag_diffs\n";
+#print "Files match\n";
 
 sub setup {
   my %opts;
   GetOptions( 'h|help' => \$opts{'h'},
               'm|man' => \$opts{'m'},
               'a|bam_a=s' => \$opts{'a'},
               'b|bam_b=s' => \$opts{'b'},
+              'c|count' => \$opts{'c'},
               's|skip' => \$opts{'s'},
   ) or pod2usage(2);
 
@@ -95,19 +111,19 @@ sub setup {
 
   my $defined = 0;
   for(keys %opts) { $defined++ if(defined $opts{$_}); }
-  if(@ARGV == 2) {
+  if($defined == 0 && @ARGV == 2) {
     $opts{'a'} = shift @ARGV;
     $opts{'b'} = shift @ARGV;
     $defined = 2;
   }
-  pod2usage(-msg  => "\nERROR: Options 'a' & 'b' must be defined.\n", -verbose => 1,  -output => \*STDERR) unless($defined == 2);
+  pod2usage(-msg  => "\nERROR: Options 'a' & 'b' must be defined.\n", -verbose => 1,  -output => \*STDERR) unless($defined >= 2);
 
   die pod2usage(-msg  => "\nERROR: bam_a and bam_b cannot be the same file\n", -verbose => 1,  -output => \*STDERR) if($opts{'a'} eq $opts{'b'});
 
   PCAP::Cli::file_for_reading('bam_a', $opts{'a'});
   PCAP::Cli::file_for_reading('bam_a', $opts{'b'});
 
-  return ($opts{'a'}, $opts{'b'}, $opts{'s'});
+  return ($opts{'a'}, $opts{'b'}, $opts{'s'}, $opts{'c'});
 }
 
 __END__
@@ -134,6 +150,7 @@ =head1 SYNOPSIS
     -bam_b    -b    The second BAM file.
 
   Other:
+    -count    -c    Count flag differences
     -skipz    -s    Don't include reads with MAPQ=0 in comparison
     -help     -h    Brief help message.
     -man      -m    Full documentation.

diff --git a/docs.tar.gz b/docs.tar.gz
diff --git a/lib/PCAP.pm b/lib/PCAP.pm
@@ -49,9 +49,10 @@ const my %UPGRADE_PATH => ( '0.1.0'  => 'biobambam,bwa,samtools',
                             '1.1.0'  => 'biobambam,bwa,samtools',
                             '1.1.1'  => 'biobambam,bwa,samtools',
                             '1.1.2'  => 'biobambam,bwa,samtools',
-                            '1.2.0'  => '', # if later versions have new versions then all preceding need that tool listing
-                            '1.2.1'  => '',
-                            '1.2.2'  => '',
+                            '1.2.0'  => 'biobambam', # if later versions have new versions then all preceding need that tool listing
+                            '1.2.1'  => 'biobambam',
+                            '1.2.2'  => 'biobambam',
+                            '1.3.0'  => '',
                           );
 
 sub license {