Copy SDRF functionality to output dir (#22)

* Initial commit for copy SDRF functionality * Complete copy part * Missing arg for sdrf copy * Fix issues * Missing curly! * Test copy of SDRF
ebi-gene-expression-group · Feb 23, 2021 · f9b8677 · f9b8677
1 parent 963ea06
commit f9b8677
Show file tree

Hide file tree

Showing 2 changed files with 35 additions and 3 deletions.
diff --git a/atlas-experiment-metadata-test.bats b/atlas-experiment-metadata-test.bats
@@ -12,6 +12,7 @@ setup() {
     implicit_pl_out_dir='implicit_pl'
     explicit_pl_out="${explicit_pl_out_dir}/E-MTAB-6077.condensed-sdrf.tsv"
     implicit_pl_out="${implicit_pl_out_dir}/E-MTAB-6077.condensed-sdrf.tsv"
+    implicit_pl_output_sdrf="${implicit_pl_out_dir}/${test_exp_acc}.sdrf.txt"
     explicit_sc_sh_out_dir='explicit_sc_sh'
     implicit_sc_sh_out_dir='implicit_sc_sh'
     zooma_exclusions="test_data/zooma_exclusions.yml"
@@ -32,15 +33,16 @@ setup() {
     [ -f "$explicit_pl_out" ]
 }
 
-@test "Test single-cell condense perl script with implicit IDF" {
+@test "Test single-cell condense perl script with implicit IDF and SDRF copy" {
     if [ -f "$implicit_pl_out" ]; then
         skip "Output from implicit condense pl exists"
     fi
 
-    run mkdir -p $implicit_pl_out_dir && env ATLAS_PROD=$test_data_dir $condense_pl -sc -e $test_exp_acc -o $implicit_pl_out_dir
+    run mkdir -p $implicit_pl_out_dir && env ATLAS_PROD=$test_data_dir $condense_pl -sc -s -e $test_exp_acc -o $implicit_pl_out_dir
 
     [ "$status" -eq 0 ]
     [ -f "$implicit_pl_out" ]
+    [ -f "$implicit_pl_output_sdrf" ]
 }
 
 @test "Test single-cell condense wrapper with explicit IDF" {

diff --git a/condense_sdrf.pl b/condense_sdrf.pl
@@ -45,6 +45,10 @@ =head1 OPTIONS
 
 Optional. Copy IDF file from ArrayExpress load directory to output directory.
 
+=item -s --copySDRF
+
+Optional. Copy SDRF file from the IDF determined location to the output directory.
+
 =item -o --outdir
 
 Optional. Destination directory for output file(s). Will use current working
@@ -105,6 +109,7 @@ =head1 AUTHOR
 use Atlas::ZoomaClient::MappingResult;
 use Atlas::AtlasConfig::Reader qw( parseAtlasFactors );
 use File::Basename;
+use File::Spec;
 use Data::Dumper;
 
 $| = 1;
@@ -164,8 +169,13 @@ =head1 AUTHOR
         ignore_datafiles    => 1
     });
 
-my $magetab = $reader->parse;
+my ($investigation, $magetab) = $reader->parse;
 $logger->info( "Successfully read MAGETAB." );
+
+if( $args->{ "copySDRF" } ) {
+    copy_sdrf_to_output_dir($investigation, $args->{ "output_directory" }, $idfFile);
+}
+
 $logger->info( "Merging technical replicates if available.") if( $args->{ "mergeTechReplicates" } );
 my $atlasAssays = create_all_atlas_assays( $magetab, $args->{ "mergeTechReplicates" } );
 $logger->debug( Dumper( $atlasAssays ) );
@@ -247,6 +257,7 @@ sub parse_args {
         "z|zooma"           => \$args{ "zooma" },
         "x|zoomaExclusions=s" => \$args{ "zooma_exclusions_path" },
         "i|idf"             => \$args{ "idf" },
+        "s|copySDRF"        => \$args{"copySDRF"},
         "f|factors=s"       => \$args{ "factors_file" },
         "b|bioreps"         => \$args{ "bioreps" },
         "d|debug"           => \$args{ "debug" },
@@ -308,6 +319,25 @@ sub copy_idf_from_ae {
     }
 }
 
+sub copy_sdrf_to_output_dir {
+    my ( $investigation, $output_dir, $idf_abs_path ) = @_;
+    foreach my $sdrf ( @{ $investigation->get_sdrfs() } ) {
+        my $filename = $sdrf->get_uri()->file();
+        if( !File::Spec->file_name_is_absolute( $filename ) ) {
+            # append IDF path
+            my $dir = dirname($idf_abs_path);
+            $filename = File::Spec->catfile( $dir, $filename );
+        } 
+        `cp $filename $output_dir`;
+        unless( $? ) {
+            $logger->info( "Successfully copied SDRF." );
+        }
+        else {
+            $logger->logdie( "Could not copy SDRF: $!" );
+        }
+    }
+}
+
 sub create_all_atlas_assays {
     my ( $magetab, $mergeTechReplicates ) = @_;
     # Create a relaxed AtlasAssayFactory.