Pacbio Run Analysis

add build from analysis directory
genome · Aug 16, 2018 · 3cb78dd · 3cb78dd
1 parent 7175dbf
commit 3cb78dd
Show file tree

Hide file tree

Showing 2 changed files with 75 additions and 9 deletions.
diff --git a/lib/Pacbio/Run/AnalysisFactoryForRsii.pm b/lib/Pacbio/Run/AnalysisFactoryForRsii.pm
@@ -41,6 +41,40 @@ sub build {
     \@analyses;
 }
 
+sub build_from_analysis_directory {
+    my ($class, $directory) = @_;
+
+    die "No analysis directory given." if not $directory;
+    die "Analysis directory given does not exist!" if not -d "$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(@analysis_files);
+    $analysis;
+}
+
 sub _load_xml {
     my ($xml_file) = @_;
 

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');
 
 };