Merge dc40e51 into 7175dbf

genome · Aug 16, 2018 · c00e0dc · c00e0dc
2 parents 7175dbf + dc40e51
commit c00e0dc
Show file tree

Hide file tree

Showing 6 changed files with 172 additions and 40 deletions.
diff --git a/lib/Pacbio/Command/Assembly/BaxToBam/GenerateCommands.pm b/lib/Pacbio/Command/Assembly/BaxToBam/GenerateCommands.pm
@@ -4,7 +4,9 @@ use strict;
 use warnings 'FATAL';
 
 use IO::File;
+use List::Util;
 use Pacbio::Run;
+use Pacbio::Run::AnalysisFactoryForRsii;
 use Path::Class;
 
 class Pacbio::Command::Assembly::BaxToBam::GenerateCommands {
@@ -14,10 +16,10 @@ class Pacbio::Command::Assembly::BaxToBam::GenerateCommands {
             is => 'Text',
             doc => 'Command to fill in with BAX files and logging files. BAX files will be appended to the command. Use %LOG for the log file base name for each cell.  Example: bsub -o /my-logging-dir/%LOG bam2bax',
         },
-        run_directories => {
+        bax_sources => {
             is => 'Text',
             is_many => 1,
-            doc => 'Pacbio run directories',
+            doc => 'Pacbio run directories OR FOF of bax files.',
         },
     },
     has_optional_input => {
@@ -39,24 +41,27 @@ class Pacbio::Command::Assembly::BaxToBam::GenerateCommands {
         },
     },
     has_optional_transient => {
-        _bam_output_directory => { is => 'Path::Class::Dir', },
-        _runs => { is => 'ARRAY', },
+        _analyses => { is => 'ARRAY', },
         _commands_fh => { is => 'IO::Handle', },
+        _bam_output_directory => { is => 'Text', },
     },
     doc => 'insert missing primary contigs from haplotigs',
 };
 
 sub __init__ {
     my ($self) = @_;
 
-    my @runs;
-    for my $directory ( $self->run_directories ) {
-        push @runs, Pacbio::Run->new(
-            directory => Path::Class::dir($directory),
-            machine_type => 'rsii',
-        );
+    my @bax_sources = $self->bax_sources;
+    if ( @bax_sources == 1 and -f $bax_sources[0] ) {
+        $self->_resolve_analyses_from_bax_fof($bax_sources[0]);
     }
-    $self->_runs(\@runs);
+    elsif ( List::Util::all { -d $_ } @bax_sources ) {
+        $self->_resolve_analyses_from_runs(@bax_sources);
+    }
+    else {
+        $self->fatal_message('Can not handle mix of run directories and bax FOFs! %s', join("\n", @bax_sources));
+    }
+    $self->fatal_message("No analyses found in bax sources!\n%s", join("\n", @bax_sources)) if not @{$self->_analyses};
 
     my $commands_file = $self->commands_file;
     if ( $commands_file and $commands_file ne '-' and -s $commands_file ) {
@@ -77,26 +82,49 @@ sub __init__ {
 
 }
 
-sub execute {
-    my ($self) = @_;
+sub _resolve_analyses_from_bax_fof {
+    my ($self, $bax_fof) = @_;
 
-    $self->__init__;
+    my $fh = IO::File->new($bax_fof, 'r');
+    $self->fatal_message('Failed to open bax FOF!') if not $fh;
+    my @analysis_directories = List::MoreUtils::uniq( map { chomp; file($_)->parent->parent } $fh->getlines );
+    $fh->close;
+
+    my @analyses = map { Pacbio::Run::AnalysisFactoryForRsii->build_from_analysis_directory($_) } @analysis_directories;
+    $self->_analyses(\@analyses);
+}
+
+sub _resolve_analyses_from_runs {
+    my ($self, @runs) = @_;
 
     my $library_name = $self->library_name;
     my $regex = qr/$library_name/;
-
-    for my $run ( @{$self->_runs} ) {
-        my $analyses = $run->analyses_for_sample($regex);
-        if ( not $analyses ) {
+    my @analyses;
+    for my $directory ( $self->bax_sources ) {
+        my $run = Pacbio::Run->new(
+            directory => Path::Class::dir($directory),
+            machine_type => 'rsii',
+        );
+        my $run_analyses = $run->analyses_for_sample($regex);
+        if ( not $run_analyses ) {
             $self->warning_message("No analyses found for library name %s on run %", $library_name, $run->__name__);
             next;
         }
-        for my $analysis ( @$analyses ) {
-            my $bam = $self->_bam_output_for_analysis($analysis);
-            next if $bam and -s $bam;
-            my $cmd = $self->_bax_to_bam_command_for_analysis($analysis);
-            $self->_commands_fh->print("$cmd\n");
-        }
+        push @analyses, @$run_analyses;
+    }
+    $self->_analyses(\@analyses);
+}
+
+sub execute {
+    my ($self) = @_;
+
+    $self->__init__;
+
+    for my $analysis ( @{$self->_analyses} ) {
+        my $bam = $self->_bam_output_for_analysis($analysis);
+        next if $bam and -s $bam;
+        my $cmd = $self->_bax_to_bam_command_for_analysis($analysis);
+        $self->_commands_fh->print("$cmd\n");
     }
 
     1;

diff --git a/lib/Pacbio/Run/AnalysisFactoryForRsii.pm b/lib/Pacbio/Run/AnalysisFactoryForRsii.pm
@@ -14,7 +14,7 @@ sub build {
     my ($class, $directory) = @_;
 
     die "No run directory given." if not $directory;
-    die "Run directory given does not exist!" if not -d "$directory";
+    die "Run directory given does not exist! $directory" if not -d "$directory";
 
     my (@analyses);
     find(
@@ -41,6 +41,40 @@ sub build {
     \@analyses;
 }
 
+sub build_from_analysis_directory {
+    my ($class, $directory) = @_;
+
+    die "No analysis directory given." if not $directory;
+    die "Analysis directory given does not exist! $directory" if not -d "$directory";
+
+    my ($metadata_xml_file, @analysis_files);
+    find(
+        {
+            wanted => sub{
+                if ( /metadata\.xml$/) {
+                    die "Found more than one metadata XML in $directory" if $metadata_xml_file;
+                    $metadata_xml_file = $File::Find::name;
+                }
+                elsif ( $File::Find::dir =~ /Analysis_Results/ and /\.h5$/ ) {
+                    push @analysis_files, $File::Find::name;
+                }
+            },
+        },
+        glob($directory->file('*')->stringify),
+    );
+
+    die "Failed to find analysis metadata xml in directory: $directory" if !$metadata_xml_file;
+    die "Failed to find analysis files in directory: $directory" if !@analysis_files;
+
+    my $xml_info = _load_xml($metadata_xml_file);
+    my $analysis = Pacbio::Run::Analysis->new(
+        metadata_xml_file => file($metadata_xml_file),
+        %$xml_info,
+    );
+    $analysis->add_analysis_files( map { file($_) } @analysis_files );
+    $analysis;
+}
+
 sub _load_xml {
     my ($xml_file) = @_;
 

diff --git a/t/Pacbio-Command-Assembly-BaxToBam-GenerateCommands.t b/t/Pacbio-Command-Assembly-BaxToBam-GenerateCommands.t
@@ -5,9 +5,10 @@ use warnings 'FATAL';
 
 use TenxTestEnv;
 
+use File::Temp;
 use File::Slurp;
 use Test::Exception;
-use Test::More tests => 4;
+use Test::More tests => 5;
 
 my %test = ( class => 'Pacbio::Command::Assembly::BaxToBam::GenerateCommands', );
 subtest 'setup' => sub{
@@ -24,11 +25,11 @@ subtest 'setup' => sub{
 
 };
 
-subtest 'execute' => sub{
+subtest 'execute with runs' => sub{
     plan tests => 4;
 
     my $cmd = $test{class}->create(
-        run_directories => [ $test{data_dir}->file('6U00E3')->stringify, ],
+        bax_sources => [ $test{data_dir}->file('6U00E3')->stringify, ],
         bam_to_bax_command => 'bsub -q long -o %LOG bam2bax',
     );
     ok($cmd, 'create command');
@@ -45,12 +46,12 @@ subtest 'execute' => sub{
 
 };
 
-subtest 'execute with library name' => sub{
+subtest 'execute with runs and library name' => sub{
     plan tests => 4;
 
     my $library_name = 'EEAI';
     my $cmd = $test{class}->create(
-        run_directories => [ $test{data_dir}->file('6U00E3')->stringify, ],
+        bax_sources => [ $test{data_dir}->file('6U00E3')->stringify, ],
         bam_to_bax_command => 'bsub -q long -o %LOG bam2bax',
         library_name => $library_name,
     );
@@ -68,11 +69,40 @@ subtest 'execute with library name' => sub{
 
 };
 
+subtest 'execute with bax fof' => sub{
+    plan tests => 4;
+
+    my $bax_fof_contents = File::Slurp::slurp($test{data_dir}->file('bax.fof')->stringify);
+    my $base_test_data_dir = TenxTestEnv::test_data_directory();
+    $bax_fof_contents =~ s/\%TDD/$base_test_data_dir/g;
+
+    my ($fh, $bax_fof) = File::Temp::tempfile();
+    $fh->print($bax_fof_contents);
+    $fh->close;
+
+    my $cmd = $test{class}->create(
+        bax_sources => [ $bax_fof ],
+        bam_to_bax_command => 'bsub -q long -o %LOG bam2bax',
+    );
+    ok($cmd, 'create command');
+
+    my $output;
+    open local(*STDOUT), '>', \$output or die $!;
+    lives_ok(sub{ $cmd->execute }, 'execute');
+    ok($cmd->result, 'command result');
+
+    my $expected_output = File::Slurp::slurp( $test{data_dir}->file('expected.bax-fof.out')->stringify );
+    $base_test_data_dir = TenxTestEnv::test_data_directory();
+    $expected_output =~ s/\%TDD/$base_test_data_dir/g;
+    is($output, $expected_output, 'output commands matches');
+
+};
+
 subtest 'execute with some bams completed' => sub{
     plan tests => 4;
 
     my $cmd = $test{class}->create(
-        run_directories => [ $test{data_dir}->file('6U00E3')->stringify, ],
+        bax_sources => [ $test{data_dir}->file('6U00E3')->stringify, ],
         bam_to_bax_command => 'bsub -q long -o %LOG bam2bax',
         bam_output_directory => $test{data_dir}->stringify,
     );

diff --git a/t/Pacbio-Run-AnalysisFactoryForRsii.t b/t/Pacbio-Run-AnalysisFactoryForRsii.t
@@ -6,29 +6,61 @@ use warnings 'FATAL';
 use TenxTestEnv;
 
 use Path::Class;
-use Test::More tests => 1;
+use Test::More tests => 3;
 use Test::Exception;
 
 my %setup = ( class => 'Pacbio::Run::AnalysisFactoryForRsii', );
-subtest 'new' => sub{
-    plan tests => 11;
+subtest 'setup' => sub{
+    plan tests => 3;
 
     use_ok($setup{class}) or die;
 
-    throws_ok(sub{ $setup{class}->build; }, qr/No run directory given/, 'new fails w/o directory');
-    throws_ok(sub{ $setup{class}->build('blah'); }, qr/Run directory given does not exist/, 'new fails w/ non existing directory');
+    $setup{run_dir} = TenxTestEnv::test_data_directory_for_class('Pacbio::Run')->subdir('6U00FA');
+    ok(-d $setup{run_dir}->stringify, 'run data dir exists');
 
-    my $directory = dir( TenxTestEnv::test_data_directory_for_class('Pacbio::Run') )->subdir('6U00FA');
-    ok(-d "$directory", "example run directory exists");
+    my $subdir = $setup{run_dir}->subdir('A01_1')->subdir('Analysis_Results');
+    my @afiles = map {
+        $subdir->file( sprintf('m160819_231415_00116_c101036512550000001823251411171640_s1_p0.%d.bax.h5', $_) )
+    } (qw/ 1 2 3 /);
+    push @afiles, $subdir->file('m160819_231415_00116_c101036512550000001823251411171640_s1_p0.bas.h5');
+    $setup{A01_1_analysis_files} = \@afiles;
+    is(@{$setup{A01_1_analysis_files}}, 4, 'run analysis files');
 
-    my $analyses = $setup{class}->build($directory);
+};
+
+subtest 'build' => sub{
+    plan tests => 10;
+
+    throws_ok(sub{ $setup{class}->build; }, qr/No run directory given/, 'fails w/o directory');
+    throws_ok(sub{ $setup{class}->build('blah'); }, qr/Run directory given does not exist/, 'fails w/ non existing directory');
+
+    my $analyses = $setup{class}->build($setup{run_dir});
     is(@$analyses, 10, 'built the correct number of analyses');
-    is($analyses->[0]->metadata_xml_file, $directory->subdir('A01_1')->file('m160819_231415_00116_c101036512550000001823251411171640_s1_p0.metadata.xml'), 'metadata_xml_file');
+    is($analyses->[0]->metadata_xml_file, $setup{run_dir}->subdir('A01_1')->file('m160819_231415_00116_c101036512550000001823251411171640_s1_p0.metadata.xml'), 'metadata_xml_file');
     is($analyses->[0]->sample_name, 'NA19434_4808o3_lib1_50pM', 'sample_name');
     is($analyses->[0]->library_name, 'NA19434_4808o3_lib1_50pM_A1', 'library_name');
     is($analyses->[0]->plate_id, '6U00FA', 'plate_id');
     is($analyses->[0]->version, '2.3.0.3.154799', 'version');
     is($analyses->[0]->well, 'A01', 'well');
+    is_deeply($analyses->[0]->analysis_files, $setup{A01_1_analysis_files}, 'analysis files');
+
+};
+
+subtest 'build from analysis directory' => sub{
+    plan tests => 9;
+
+    throws_ok(sub{ $setup{class}->build_from_analysis_directory; }, qr/No analysis directory given/, 'fails w/o directory');
+    throws_ok(sub{ $setup{class}->build_from_analysis_directory('blah'); }, qr/Analysis directory given does not exist/, 'fails w/ non existing directory');
+    throws_ok(sub{ $setup{class}->build_from_analysis_directory($setup{run_dir}->subdir('A01_1')->subdir('Analysis_Results')); }, qr/Failed to find analysis metadata xml in/, 'fails w/ when no analysis found');
+
+    my $directory = $setup{run_dir}->subdir('A01_1');
+    my $analysis = $setup{class}->build_from_analysis_directory($directory);
+    is($analysis->metadata_xml_file, $directory->file('m160819_231415_00116_c101036512550000001823251411171640_s1_p0.metadata.xml'), 'metadata_xml_file');
+    is($analysis->sample_name, 'NA19434_4808o3_lib1_50pM', 'sample_name');
+    is($analysis->library_name, 'NA19434_4808o3_lib1_50pM_A1', 'library_name');
+    is($analysis->plate_id, '6U00FA', 'plate_id');
+    is($analysis->version, '2.3.0.3.154799', 'version');
+    is($analysis->well, 'A01', 'well');
 
 };
 

diff --git a/t/data/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/bax.fof b/t/data/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/bax.fof
@@ -0,0 +1,6 @@
+%TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/B02_1/Analysis_Results/m160613_070741_00116_c100976392550000001823226708101601_s1_p0.1.bax.h5
+%TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/B02_1/Analysis_Results/m160613_070741_00116_c100976392550000001823226708101601_s1_p0.2.bax.h5
+%TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/B02_1/Analysis_Results/m160613_070741_00116_c100976392550000001823226708101601_s1_p0.3.bax.h5
+%TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/E01_1/Analysis_Results/m160611_231419_00116_c100976122550000001823226708101634_s1_p0.1.bax.h5
+%TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/E01_1/Analysis_Results/m160611_231419_00116_c100976122550000001823226708101634_s1_p0.2.bax.h5
+%TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/E01_1/Analysis_Results/m160611_231419_00116_c100976122550000001823226708101634_s1_p0.3.bax.h5
diff --git a/t/data/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/expected.bax-fof.out b/t/data/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/expected.bax-fof.out
@@ -0,0 +1,2 @@
+bsub -q long -o 6U00E3_B02.out bam2bax %TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/B02_1/Analysis_Results/m160613_070741_00116_c100976392550000001823226708101601_s1_p0.1.bax.h5 %TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/B02_1/Analysis_Results/m160613_070741_00116_c100976392550000001823226708101601_s1_p0.2.bax.h5 %TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/B02_1/Analysis_Results/m160613_070741_00116_c100976392550000001823226708101601_s1_p0.3.bax.h5
+bsub -q long -o 6U00E3_E01.out bam2bax %TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/E01_1/Analysis_Results/m160611_231419_00116_c100976122550000001823226708101634_s1_p0.1.bax.h5 %TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/E01_1/Analysis_Results/m160611_231419_00116_c100976122550000001823226708101634_s1_p0.2.bax.h5 %TDD/Pacbio-Command-Assembly-BaxToBam-GenerateCommands/6U00E3/E01_1/Analysis_Results/m160611_231419_00116_c100976122550000001823226708101634_s1_p0.3.bax.h5