Merge pull request #1820 from Clinical-Genomics/feature/rename_star_f…

…usion_out feat(star_fusion): rename outfile
Clinical-Genomics · Mar 1, 2021 · 9a4246c · 9a4246c
2 parents f3ecff6 + 0e84958
commit 9a4246c
Show file tree

Hide file tree

Showing 7 changed files with 198 additions and 52 deletions.
diff --git a/definitions/rd_rna_parameters.yaml b/definitions/rd_rna_parameters.yaml
@@ -397,7 +397,7 @@ star_fusion:
     - mip
   data_type: SCALAR
   default: 1
-  file_tag: _fusion
+  file_tag: _starfusion
   outfile_suffix: ".tsv"
   program_executables:
     - gtf_file_to_feature_seqs.pl

diff --git a/lib/MIP/File_info.pm b/lib/MIP/File_info.pm
@@ -33,6 +33,7 @@ BEGIN {
       get_io_files
       get_is_sample_files_compressed
       get_merged_infile_prefix
+      get_sample_fastq_file_lanes
       get_sample_file_attribute
       get_sampling_fastq_files
       parse_file_compression_features
@@ -685,6 +686,40 @@ sub get_sample_file_attribute {
     return $stored_attribute;
 }
 
+sub get_sample_fastq_file_lanes {
+
+## Function : Get sample fastq file lanes
+## Returns  :
+## Arguments: $file_info_href => File info hash {REF}
+##          : $sample_id      => Sample id
+
+    my ($arg_href) = @_;
+
+    ## Flatten argument(s)
+    my $file_info_href;
+    my $sample_id;
+
+    my $tmpl = {
+        file_info_href => {
+            default     => {},
+            defined     => 1,
+            required    => 1,
+            store       => \$file_info_href,
+            strict_type => 1,
+        },
+        sample_id => {
+            defined     => 1,
+            required    => 1,
+            store       => \$sample_id,
+            strict_type => 1,
+        },
+    };
+
+    check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
+
+    return @{ $file_info_href->{$sample_id}{lanes} };
+}
+
 sub get_sampling_fastq_files {
 
 ## Function : Get sample fastq files. Either single-end, paired-end or interleaved

diff --git a/lib/MIP/Recipes/Analysis/Arriba.pm b/lib/MIP/Recipes/Analysis/Arriba.pm
@@ -132,7 +132,8 @@ sub analysis_arriba {
 
     check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
 
-    use MIP::File_info qw{ get_io_files get_sample_file_attribute parse_io_outfiles };
+    use MIP::File_info
+      qw{ get_io_files get_sample_fastq_file_lanes get_sample_file_attribute parse_io_outfiles };
     use MIP::Program::Gnu::Coreutils qw{ gnu_rm gnu_tee };
     use MIP::Processmanagement::Processes qw{ submit_recipe };
     use MIP::Program::Arriba qw{ arriba draw_fusions };
@@ -165,7 +166,6 @@ sub analysis_arriba {
     );
     my @infile_paths = @{ $io{in}{file_paths} };
 
-## Build outfile_paths
     my %recipe = parse_recipe_prerequisites(
         {
             active_parameter_href => $active_parameter_href,
@@ -174,29 +174,34 @@ sub analysis_arriba {
         }
     );
 
-    my $outsample_directory =
-      catdir( $active_parameter_href->{outdata_dir}, $sample_id, $recipe_name );
-    my $lanes_id            = join $EMPTY_STR, @{ $file_info_href->{$sample_id}{lanes} };
-    my $outfile_tag         = $file_info_href->{$sample_id}{$recipe_name}{file_tag};
-    my $outfile_suffix      = $recipe{outfile_suffix};
-    my $outfile_path_prefix = catfile( $outsample_directory,
-        $sample_id . $UNDERSCORE . q{lanes} . $UNDERSCORE . $lanes_id . $outfile_tag );
-
+    ## Join infile lanes
+    my $lanes_id = join $EMPTY_STR,
+      get_sample_fastq_file_lanes(
+        {
+            file_info_href => $file_info_href,
+            sample_id      => $sample_id,
+        }
+      );
     %io = (
         %io,
         parse_io_outfiles(
             {
-                chain_id       => $recipe{job_id_chain},
+                chain_id               => $recipe{job_id_chain},
+                file_info_href         => $file_info_href,
+                file_name_prefixes_ref =>
+                  [ $sample_id . $UNDERSCORE . q{lanes} . $UNDERSCORE . $lanes_id ],
                 id             => $sample_id,
-                file_info_href => $file_info_href,
-                file_paths_ref => [ $outfile_path_prefix . $outfile_suffix ],
+                outdata_dir    => $active_parameter_href->{outdata_dir},
                 parameter_href => $parameter_href,
                 recipe_name    => $recipe_name,
             }
         )
     );
-    my $outfile_name = ${ $io{out}{file_names} }[0];
-    my $outfile_path = $io{out}{file_path};
+    my $outdir_path         = $io{out}{dir_path};
+    my $outfile_name        = ${ $io{out}{file_names} }[0];
+    my $outfile_path        = $io{out}{file_path};
+    my $outfile_path_prefix = $io{out}{file_path_prefix};
+    my $outfile_suffix      = $io{out}{file_suffix};
 
     my $use_sample_id_as_display_name =
       $active_parameter_href->{arriba_use_sample_id_as_display_name};
@@ -394,8 +399,8 @@ sub analysis_arriba {
     );
     if ( $sample_display_name and not $use_sample_id_as_display_name ) {
 
-        $report_path = catfile( $outsample_directory,
-            $sample_display_name . $UNDERSCORE . q{arriba_fusions.pdf} );
+        $report_path =
+          catfile( $outdir_path, $sample_display_name . $UNDERSCORE . q{arriba_fusions.pdf} );
     }
     draw_fusions(
         {

diff --git a/lib/MIP/Recipes/Analysis/Star_aln.pm b/lib/MIP/Recipes/Analysis/Star_aln.pm
@@ -126,7 +126,8 @@ sub analysis_star_aln {
 
     check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
 
-    use MIP::File_info qw{ get_io_files get_sample_file_attribute parse_io_outfiles };
+    use MIP::File_info
+      qw{ get_io_files get_sample_fastq_file_lanes get_sample_file_attribute parse_io_outfiles };
     use MIP::Program::Gnu::Coreutils qw{ gnu_mv gnu_rm };
     use MIP::Program::Samtools qw{ samtools_index };
     use MIP::Program::Star qw{ star_aln };
@@ -165,29 +166,33 @@ sub analysis_star_aln {
         }
     );
 
-    my $outsample_directory =
-      catdir( $active_parameter_href->{outdata_dir}, $sample_id, $recipe_name );
-    my $lanes_id            = join $EMPTY_STR, @{ $file_info_href->{$sample_id}{lanes} };
-    my $outfile_tag         = $file_info_href->{$sample_id}{$recipe_name}{file_tag};
-    my $outfile_suffix      = $recipe{outfile_suffix};
-    my $outfile_path_prefix = catfile( $outsample_directory,
-        $sample_id . $UNDERSCORE . q{lanes} . $UNDERSCORE . $lanes_id . $outfile_tag );
-
+    ## Join infile lanes
+    my $lanes_id = join $EMPTY_STR,
+      get_sample_fastq_file_lanes(
+        {
+            file_info_href => $file_info_href,
+            sample_id      => $sample_id,
+        }
+      );
     %io = (
         %io,
         parse_io_outfiles(
             {
-                chain_id       => $recipe{job_id_chain},
+                chain_id               => $recipe{job_id_chain},
+                file_info_href         => $file_info_href,
+                file_name_prefixes_ref =>
+                  [ $sample_id . $UNDERSCORE . q{lanes} . $UNDERSCORE . $lanes_id ],
                 id             => $sample_id,
-                file_info_href => $file_info_href,
-                file_paths_ref => [ $outfile_path_prefix . $outfile_suffix ],
+                outdata_dir    => $active_parameter_href->{outdata_dir},
                 parameter_href => $parameter_href,
                 recipe_name    => $recipe_name,
             }
         )
     );
-    my $outfile_name = ${ $io{out}{file_names} }[0];
-    my $outfile_path = $io{out}{file_path};
+    my $outfile_name        = ${ $io{out}{file_names} }[0];
+    my $outfile_path        = $io{out}{file_path};
+    my $outfile_path_prefix = $io{out}{file_path_prefix};
+    my $outfile_suffix      = $io{out}{file_suffix};
 
     ## Filehandles
     # Create anonymous filehandle

diff --git a/lib/MIP/Recipes/Analysis/Star_fusion.pm b/lib/MIP/Recipes/Analysis/Star_fusion.pm
@@ -16,7 +16,7 @@ use autodie qw{ :all };
 use Readonly;
 
 ## MIPs lib/
-use MIP::Constants qw{ $ASTERISK $LOG_NAME $NEWLINE $UNDERSCORE };
+use MIP::Constants qw{ $ASTERISK $EMPTY_STR $LOG_NAME $NEWLINE $UNDERSCORE };
 
 BEGIN {
 
@@ -127,8 +127,8 @@ sub analysis_star_fusion {
     check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
 
     use MIP::File::Format::Star_fusion qw{ create_star_fusion_sample_file };
-    use MIP::File_info qw{ get_io_files parse_io_outfiles };
-    use MIP::Program::Gnu::Coreutils qw{ gnu_cp };
+    use MIP::File_info qw{ get_io_files parse_io_outfiles get_sample_fastq_file_lanes };
+    use MIP::Program::Gnu::Coreutils qw{ gnu_mv gnu_rm };
     use MIP::Program::Star_fusion qw{ star_fusion };
     use MIP::Processmanagement::Processes qw{ submit_recipe };
     use MIP::Recipe qw{ parse_recipe_prerequisites };
@@ -165,23 +165,33 @@ sub analysis_star_fusion {
             recipe_name           => $recipe_name,
         }
     );
-    my $outdir_path    = catdir( $active_parameter_href->{outdata_dir}, $sample_id, $recipe_name );
-    my $outsample_name = $STAR_FUSION_PREFIX . $recipe{outfile_suffix};
-    my @file_paths     = catfile( $outdir_path, $outsample_name );
 
+    ## Join infile lanes
+    my $lanes_id = join $EMPTY_STR,
+      get_sample_fastq_file_lanes(
+        {
+            file_info_href => $file_info_href,
+            sample_id      => $sample_id,
+        }
+      );
     %io = (
         %io,
         parse_io_outfiles(
             {
-                chain_id       => $recipe{job_id_chain},
+                chain_id               => $recipe{job_id_chain},
+                file_info_href         => $file_info_href,
+                file_name_prefixes_ref =>
+                  [ $sample_id . $UNDERSCORE . q{lanes} . $UNDERSCORE . $lanes_id ],
                 id             => $sample_id,
-                file_info_href => $file_info_href,
-                file_paths_ref => \@file_paths,
+                outdata_dir    => $active_parameter_href->{outdata_dir},
                 parameter_href => $parameter_href,
                 recipe_name    => $recipe_name,
             }
         )
     );
+    my $outdir_path    = $io{out}{dir_path};
+    my $outfile_path   = $io{out}{file_path};
+    my $outfile_suffix = $io{out}{file_suffix};
 
     ## Filehandles
     # Create anonymous filehandle
@@ -236,14 +246,40 @@ sub analysis_star_fusion {
     );
     say {$filehandle} $NEWLINE;
 
-    ## Close filehandleS
+    say {$filehandle} q{## Rename outfile};
+    my $star_fusion_outfile_path = catfile( $outdir_path, $STAR_FUSION_PREFIX . $outfile_suffix );
+    gnu_mv(
+        {
+            filehandle   => $filehandle,
+            infile_path  => $star_fusion_outfile_path,
+            outfile_path => $outfile_path,
+        }
+    );
+    say {$filehandle} $NEWLINE;
+
+    say {$filehandle} q{## Remove intermediary files};
+  FILE_TAG:
+    foreach my $file_tag (qw{ _STARgenome _STARpass1 }) {
+
+        gnu_rm(
+            {
+                filehandle  => $filehandle,
+                force       => 1,
+                infile_path => catdir( $outdir_path, $file_tag ),
+                recursive   => 1,
+            }
+        );
+        print {$filehandle} $NEWLINE;
+    }
+
+    ## Close filehandle
     close $filehandle or $log->logcroak(q{Could not close filehandle});
 
     if ( $recipe{mode} == 1 ) {
         ## Collect QC metadata info for later use
         set_recipe_outfile_in_sample_info(
             {
-                path             => $file_paths[0],
+                path             => $outfile_path,
                 recipe_name      => $recipe_name,
                 sample_id        => $sample_id,
                 sample_info_href => $sample_info_href,
@@ -254,7 +290,7 @@ sub analysis_star_fusion {
             {
                 format           => q{meta},
                 id               => $sample_id,
-                path             => $file_paths[0],
+                path             => $outfile_path,
                 recipe_name      => $recipe_name,
                 sample_info_href => $sample_info_href,
             }

diff --git a/t/analysis_star_fusion.t b/t/analysis_star_fusion.t
@@ -62,18 +62,20 @@ my %active_parameter = test_mip_hashes(
 $active_parameter{$recipe_name}                     = 1;
 $active_parameter{recipe_core_number}{$recipe_name} = 1;
 $active_parameter{recipe_time}{$recipe_name}        = 1;
+$active_parameter{star_fusion_genome_lib_dir}       = q{a_dir};
+$active_parameter{star_fusion_reference_genome}     = q{a_reference_genome};
 my $sample_id = $active_parameter{sample_ids}[0];
-$active_parameter{star_fusion_genome_lib_dir}   = q{a_dir};
-$active_parameter{star_fusion_reference_genome} = q{a_reference_genome};
 
 my %file_info = test_mip_hashes(
     {
         mip_hash_name => q{file_info},
         recipe_name   => $recipe_name,
     }
 );
-
 $file_info{star_fusion_reference_genome} = [qw{ _star_fusion_genome_dir }];
+@{ $file_info{$sample_id}{lanes} } = ( 1, 2 );
+$file_info{$sample_id}{$recipe_name}{file_tag} = q{starfusion};
+
 my %job_id;
 my %parameter = test_mip_hashes(
     {
@@ -84,11 +86,11 @@ my %parameter = test_mip_hashes(
 
 test_add_io_for_recipe(
     {
-        file_info_href    => \%file_info,
-        id                => $sample_id,
-        parameter_href    => \%parameter,
-        recipe_name       => $recipe_name,
-        step              => q{bam},
+        file_info_href => \%file_info,
+        id             => $sample_id,
+        parameter_href => \%parameter,
+        recipe_name    => $recipe_name,
+        step           => q{bam},
     }
 );