Merge pull request #594 from wtsi-npg/devel

pull from devel to master to create release 91.8.0

Author: mgcam

Changes

@@ -1,8 +1,13 @@
 LIST OF CHANGES
 
+release 91.8.0
+ - BotSeq renamed Duplex-Seq
+ - added new functions to long_info for BotSeq analysis
+ - added support for NovaSeq v1.5 reagents, i5 index sequenced in opposite direction
+
 release 91.7.0
  - remove the template for an unused tracking server url - run/summary -
-   and all underlying code to remove obsolete sql queries used by that code
+   and all underlying code to remove obsolete SQL queries used by that code
 
 release 91.6.0
  - enforce custom type NpgTrackingReadableFile to check that the

MANIFEST

@@ -968,9 +968,12 @@ t/data/run_info/runInfo.miseq.xml
 t/data/run_info/runInfo.nextseq.xml
 t/data/run_info/runInfo.novaseq.xml
 t/data/run_info/runInfo.novaseq.xp.xml
+t/data/run_info/runInfo.novaseq.xp.v1.5.xml
 t/data/run_params/RunParameters.miniseq.xml
 t/data/run_params/RunParameters.nextseq.xml
 t/data/run_params/RunParameters.novaseq.xml
+t/data/run_params/RunParameters.novaseq.xp.xml
+t/data/run_params/RunParameters.novaseq.xp.v1.5.xml
 t/data/run_params/runParameters.hiseq.rr.single.xml
 t/data/run_params/runParameters.hiseq.rr.twoind.xml
 t/data/run_params/runParameters.hiseq.rr.xml

lib/npg_tracking/illumina/run/long_info.pm

@@ -8,10 +8,12 @@ use IO::All;
 use File::Spec;
 use XML::LibXML;
 use Try::Tiny;
+use Readonly;
 
 requires qw{runfolder_path};
 
 our $VERSION = '0';
+Readonly::Scalar my $NOVASEQ_I5FLIP_REAGENT_VER              => 3;
 
 =head1 NAME
 
@@ -104,6 +106,25 @@ sub _build_index_length {
   return $end - $start + 1;
 }
 
+=head2 is_dual_index
+
+Boolean determines if the run has a second index read, this can be set on object construction
+
+  my $bIsDualIndex = $class->is_dual_index();
+
+=cut
+
+has q{is_dual_index} => (
+  isa           => q{Bool},
+  is            => q{ro},
+  lazy_build    => 1,
+  documentation => q{This run is a paired end read},
+);
+sub _build_is_dual_index {
+  my $self = shift;
+  return $self->index_read2_cycle_range ? 1 : 0;
+}
+
 =head2 lane_count
 
 Number of lanes configured for this run. May be set on Construction.
@@ -227,6 +248,42 @@ has _read2_cycle_range => (
   },
 );
 
+=head2 index_read1_cycle_range
+
+First and last cycles of index read 1, or nothing returned if no index_read 1
+
+=cut
+
+has q{_index_read1_cycle_range} => (
+  traits  => ['Array'],
+  is      => 'ro',
+  isa     => 'ArrayRef[Int]',
+  default => sub { [] },
+  handles => {
+    _push_index_read1_cycle_range  => 'push',
+    index_read1_cycle_range        => 'elements',
+    has_index_read1_cycle_range    => 'count',
+  },
+);
+
+=head2 index_read2_cycle_range
+
+First and last cycles of index_read 2, or nothing returned if no index_read 2
+
+=cut
+
+has _index_read2_cycle_range => (
+  traits  => ['Array'],
+  is      => 'ro',
+  isa     => 'ArrayRef[Int]',
+  default => sub { [] },
+  handles => {
+    _push_index_read2_cycle_range => 'push',
+    index_read2_cycle_range       => 'elements',
+    has_index_read2_cycle_range   => 'count',
+  },
+);
+
 =head2 expected_cycle_count
 
 Number of cycles configured for this run and for which the output data (images or intensities or both) can be expected to be found below this folder. This number is extracted from the recipe file. It does not include the cycles for the paired read if that is performed as a separate run - the output data for that will be in a different runfolder.
@@ -409,6 +466,8 @@ foreach my $f ( qw(expected_cycle_count
                    indexing_cycle_range
                    read1_cycle_range
                    read2_cycle_range
+                   index_read1_cycle_range
+                   index_read2_cycle_range
                    tilelayout_rows
                    tilelayout_columns) ) {
    before $f => sub {
@@ -541,6 +600,20 @@ sub _build_flowcell_mode {
   return _get_single_element_text($self->_run_params(), 'FlowCellMode');
 }
 
+=head2 sbs_consumable_version
+
+=cut
+
+has q{sbs_consumable_version} => (
+  isa        => 'Str',
+  is         => 'ro',
+  lazy_build => 1,
+);
+sub _build_sbs_consumable_version{
+  my $self = shift;
+  return _get_single_element_text($self->_run_params(), 'SbsConsumableVersion')||1;
+}
+
 =head2 all_lanes_mergeable
 
 Method returns true if all lanes on the flowcell contain the
@@ -605,20 +678,24 @@ sub is_rapid_run_abovev2 {
 
 =head2 is_i5opposite
 
-A dual-indexed sequencing run on the MiniSeq, NextSeq, HiSeq 4000, or HiSeq 3000
+A dual-indexed sequencing run on the MiniSeq, NextSeq, HiSeq 4000, HiSeq 3000 or NovaSeq using v1.5 reagents
 performs the Index 2 Read after the Read 2 resynthesis step. This workflow
 requires a reverse complement of the Index 2 (i5) primer sequence compared to
 the primer sequence used on other Illumina platform, see
 https://support.illumina.com/content/dam/illumina-support/documents/documentation/system_documentation/miseq/indexed-sequencing-overview-guide-15057455-04.pdf
+For NovaSeq using v1.5 reagents the SbsConsumableVersion will be 3
 
 Method returns true if this is the case.
 
 =cut
 
 sub is_i5opposite {
   my $self = shift;
-  return ($self->is_paired_read() && ($self->platform_HiSeqX()  or $self->platform_HiSeq4000() or
-                                      $self->platform_MiniSeq() or $self->platform_NextSeq()));
+  return ($self->is_paired_read() &&
+              ($self->platform_HiSeqX()  or $self->platform_HiSeq4000() or
+               $self->platform_MiniSeq() or $self->platform_NextSeq() or
+               ($self->platform_NovaSeq() &&
+                ($self->sbs_consumable_version() >= $NOVASEQ_I5FLIP_REAGENT_VER))));
 }
 
 =head2 uses_patterned_flowcell
@@ -861,11 +938,13 @@ sub _set_values_at_end_of_read {
         if ( $rc->{start} == $end + 1 ) {
           $self->_pop_indexing_cycle_range();
           $self->_push_indexing_cycle_range( $rc->{index} );
+          $self->_push_index_read2_cycle_range( $rc->{start},$rc->{index} );
         } else {
           carp "Don't know how to deal with non adjacent indexing reads: $start,$end and $rc->{start},$rc->{index}"
         }
       } else {
         $self->_push_indexing_cycle_range( $rc->{start},$rc->{index} );
+        $self->_push_index_read1_cycle_range( $rc->{start},$rc->{index} );
       }
     } else {
       $self->_push_reads_indexed(0);

t/60-illumina-run-long_info.t

@@ -1,6 +1,6 @@
 use strict;
 use warnings;
-use Test::More tests => 59;
+use Test::More tests => 65;
 use Test::Exception;
 use Test::Deep;
 use File::Temp qw(tempdir);
@@ -28,7 +28,7 @@ package main;
 my $basedir = tempdir( CLEANUP => 1 );
 
 subtest 'retrieving information from runParameters.xml' => sub {
-  plan tests => 142;
+  plan tests => 153;
 
   my $rf = join q[/], $basedir, 'runfolder';
   mkdir $rf;
@@ -50,6 +50,7 @@ subtest 'retrieving information from runParameters.xml' => sub {
     RunParameters.nextseq.xml
     RunParameters.novaseq.xml
     RunParameters.novaseq.xp.xml
+    RunParameters.novaseq.xp.v1.5.xml
     runParameters.hiseq.rr.truseq.xml
                   /;
   my $dir = 't/data/run_params';
@@ -119,7 +120,7 @@ subtest 'retrieving information from runParameters.xml' => sub {
 };
 
 subtest 'getting i5opposite for run' => sub {
-  plan tests => 14;
+  plan tests => 15;
 
   $basedir = tempdir( CLEANUP => 1 );
   my $rf = join q[/], $basedir, 'run_info';
@@ -144,6 +145,7 @@ subtest 'getting i5opposite for run' => sub {
     'runInfo.nextseq.xml'                  => { 'rpf' => 'RunParameters', 'i5opposite' => 1 },
     'runInfo.novaseq.xml'                  => { 'rpf' => 'RunParameters', 'i5opposite' => 0 },
     'runInfo.novaseq.xp.xml'               => { 'rpf' => 'RunParameters', 'i5opposite' => 0 },
+    'runInfo.novaseq.xp.v1.5.xml'          => { 'rpf' => 'RunParameters', 'i5opposite' => 1 },
   );
 
   my $run_info_dir = 't/data/run_info';
@@ -433,6 +435,12 @@ note($long_info->runfolder_path);
   lives_and { ok( $long_info->is_indexed, 'is_indexed ok');} 'is_indexed lives and ok';
   lives_ok { $long_info = test::long_info->new({runfolder_path=>$rfpath}); } q{create test role for dual index paired};
   lives_and { ok( $long_info->is_paired_read, 'is_paired_read ok');} 'is_paired_read lives and ok';
+  lives_ok { $long_info = test::long_info->new({runfolder_path=>$rfpath}); } q{create test role for dual index paired};
+  lives_and { ok( $long_info->is_dual_index, 'is_dual_index ok');} 'is_paired_read lives and ok';
+  lives_ok { $long_info = test::long_info->new({runfolder_path=>$rfpath}); } q{create test role for dual index paired};
+  lives_and { cmp_deeply( [$long_info->index_read1_cycle_range], [152,159], 'index_read1_cycle_range matches');} 'index_read1_cycle_range lives and matches';
+  lives_ok { $long_info = test::long_info->new({runfolder_path=>$rfpath}); } q{create test role for dual index paired};
+  lives_and { cmp_deeply( [$long_info->index_read2_cycle_range], [160,167], 'index_read2_cycle_range matches');} 'index_read2_cycle_range lives and matches';
 }
 
 #and a SP flowcell