Merge branch 'hotfix/v3.0.4'

cancerit · Nov 22, 2018 · 1850681 · 1850681
2 parents 41fa2a9 + a1be7c2
commit 1850681
Show file tree

Hide file tree

Showing 6 changed files with 30 additions and 7 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,5 +1,9 @@
 # CHANGES
 
+## 3.0.4
+
+Output bug for pindel BAM/CRAM corrected.  When more than 1 chr in output files had no reads.
+
 ## 3.0.3
 
 Changes to how germline filter determined resulted in dummy germline bed file not being generated as previously.

diff --git a/perl/bin/pindel_merge_vcf_bam.pl b/perl/bin/pindel_merge_vcf_bam.pl
@@ -115,7 +115,7 @@ sub setup {
       push @{$opts{'vcf'}}, $file;
     }
     elsif($file =~ m/_([mw]t)\.sam(.gz)?$/) {
-      next if($file =~ m/^FINAL_MERGED\.sam$/); # this is transient for each type
+      next if($file =~ m/^FINAL_MERGED\.sam.gz$/); # this is transient for each type
       push @{$opts{$1}}, $file;
     }
   }

diff --git a/perl/docs.tar.gz b/perl/docs.tar.gz
diff --git a/perl/lib/Sanger/CGP/Pindel.pm b/perl/lib/Sanger/CGP/Pindel.pm
@@ -26,7 +26,7 @@ use strict;
 use Const::Fast qw(const);
 
 use base 'Exporter';
-our $VERSION = '3.0.3';
+our $VERSION = '3.0.4';
 our @EXPORT = qw($VERSION);
 
 1;
diff --git a/perl/lib/Sanger/CGP/Pindel/Implement.pm b/perl/lib/Sanger/CGP/Pindel/Implement.pm
@@ -403,7 +403,9 @@ sub fragmented_files {
   my $file_count = scalar @{$files};
   return $files if($file_count <= 100);
   warn "Extreme number of files, slow merging of $file_count files required\n";
-  my $grep_non_header = qq{zgrep -v '^$hprefix' %s >> %s};
+  my $extra = ' | gzip -c ';
+  $extra = q{} if($outfile =~ m/vcf$/);
+  my $grep_non_header = qq{zgrep -v '^$hprefix' %s $extra >> %s};
   my $header_count;
 
   my $merged_file = $base_dir.$outfile;

diff --git a/perl/lib/Sanger/CGP/Pindel/OutputGen/BamUtil.pm b/perl/lib/Sanger/CGP/Pindel/OutputGen/BamUtil.pm
@@ -42,7 +42,7 @@ const my $BAMSORT => ' inputformat=%s outputformat=%s reference=%s | tee %s | md
 
 sub sam_to_sorted_bam {
   my ($path_prefix, $base_dir, $sam_files, $as_cram, $as_csi, $ref) = @_;
-  $sam_files = Sanger::CGP::Pindel::Implement::fragmented_files($base_dir, $sam_files, '@', 'FINAL_MERGED.sam');
+  $sam_files = Sanger::CGP::Pindel::Implement::fragmented_files($base_dir, $sam_files, '@', 'FINAL_MERGED.sam.gz');
   update_header_when_no_reads($base_dir, $sam_files);
   my $aln_fmt = 'bam';
   my $idx_fmt = 'bai';
@@ -60,13 +60,24 @@ sub sam_to_sorted_bam {
   my $aln_idx = $aln_out.'.'.$idx_fmt;
   my $aln_md5 = $aln_out.'.md5';
   my $command = q{};
-  $command .= "cd $base_dir; " unless($sam_files->[0] =~ m/FINAL_MERGED[.]sam$/);
-  $command .= sprintf ' zcat %s | ', join q{ }, @{$sam_files};
+  if($sam_files->[0] =~ m/FINAL_MERGED[.]sam.gz$/) {
+    $command .= sprintf ' zcat %s | ', $sam_files->[0];
+  }
+  else {
+    my $first_file = shift @{$sam_files};
+    $command .= "cd $base_dir; ";
+    $command .= sprintf q{ (zcat %s ; zgrep -v '^@' %s) | }, $first_file, join q{ }, @{$sam_files};
+  }
+
   $command .= which('bamsort');
   $command .= sprintf $BAMSORT, 'sam', $aln_fmt, $ref, $aln_out, $aln_md5;
+
+  $command = sprintf q{bash -c "set -o pipefail; %s"}, $command;
+  warn "running: $command";
   system($command);
 
   $command = sprintf 'samtools index %s %s %s', $idx_switch, $aln_out, $aln_idx;
+  warn "running: $command";
   system($command);
 
   return 1;
@@ -75,7 +86,13 @@ sub sam_to_sorted_bam {
 sub update_header_when_no_reads {
   my ($base_dir, $sam_files) = @_;
   for my $sam_file (@{$sam_files}) {
-    my $sam = $base_dir.$sam_file;
+    my $sam;
+    if($sam_file =~ m/FINAL_MERGED[.]sam.gz$/) {
+      $sam = $sam_file
+    }
+    else {
+      $sam = $base_dir.$sam_file;
+    }
     my @header;
     my $has_records = 0;
     open my $IN, '<:gzip', $sam || die $!;