Merge pull request #1844 from Clinical-Genomics/feature/clinical_fusi…

…on_report feat(fusion): produce clinical fusion report
Clinical-Genomics · Mar 18, 2021 · f3d97a2 · f3d97a2
2 parents 4b3ddd5 + 7660092
commit f3d97a2
Show file tree

Hide file tree

Showing 21 changed files with 1,057 additions and 145 deletions.
diff --git a/definitions/rd_rna_initiation_map.yaml b/definitions/rd_rna_initiation_map.yaml
@@ -13,8 +13,10 @@ CHAIN_ALL:
     - star_fusion
     - arriba_ar
     - megafusion_ar
-    - svdb_merge_fusion
     - fusion_report
+    - PARALELL:
+      - svdb_merge_fusion
+      - merge_fusion_reports
   - CHAIN_MAIN:
     - star_aln
     - picardtools_mergesamfiles

diff --git a/definitions/rd_rna_parameters.yaml b/definitions/rd_rna_parameters.yaml
@@ -6,6 +6,7 @@ custom_default_parameters:
   default:
     - analysis_type
     - conda_path
+    - fusion_select_file
     - infile_dirs
     - install_config_file
     - pedigree_fam_file
@@ -146,6 +147,7 @@ recipe_time:
     markduplicates: 8
     megafusion_ar: 1
     multiqc_ar: 1
+    merge_fusion_reports: 1
     picardtools_collectrnaseqmetrics: 2
     picardtools_mergesamfiles: 2
     preseq_ar: 2
@@ -206,8 +208,9 @@ recipe_core_number:
     genebody_coverage: 1
     gffcompare_ar: 1
     gzip_fastq: 0
-    multiqc_ar: 1
     markduplicates: 13
+    merge_fusion_reports: 1
+    multiqc_ar: 1
     picardtools_collectrnaseqmetrics: 1
     picardtools_mergesamfiles: 13
     preseq_ar: 1
@@ -457,16 +460,16 @@ arriba_blacklist_path:
   data_type: SCALAR
   type: path
 fusion_report:
-  analysis_mode: case
+  analysis_mode: sample
   associated_recipe:
     - mip
   data_type: SCALAR
   default: 1
   file_tag: _report
   outfile_suffix: ".pdf"
   program_executables:
+    - perl
     - draw_fusions.R
-    - pdfmerger
   type: recipe
 fusion_cytoband_path:
   associated_recipe:
@@ -478,6 +481,13 @@ fusion_proteindomain_path:
     - fusion_report
   data_type: SCALAR
   type: path
+fusion_select_file:
+  associated_recipe:
+    - fusion_report
+  data_type: SCALAR
+  exists_check: file
+  type: path
+  update_path: absolute_path
 fusion_use_sample_id_as_display_name:
   associated_recipe:
     - arriba_ar
@@ -522,13 +532,24 @@ svdb_merge_fusion:
     - mip
   data_type: SCALAR
   default: 1
-  file_tag: _fusion
+  file_tag: _fusion_report
   outfile_suffix: ".vcf.gz"
   program_executables:
     - bcftools
     - svdb
   type: recipe
 ## Merge bam files
+merge_fusion_reports:
+  analysis_mode: case
+  associated_recipe:
+    - mip
+  data_type: SCALAR
+  default: 1
+  file_tag: _report
+  outfile_suffix: ".pdf"
+  program_executables:
+    - pdfmerger
+  type: recipe
 picardtools_mergesamfiles:
   analysis_mode: sample
   associated_recipe:

diff --git a/lib/MIP/Active_parameter.pm b/lib/MIP/Active_parameter.pm
@@ -2584,8 +2584,9 @@ sub set_vcfparser_outfile_counter {
 
     ## Create link
     my %vcfparser_select_file = (
-        sv_vcfparser => { sv_vcfparser_select_file => q{sv_vcfparser_outfile_count} },
-        vcfparser_ar => { vcfparser_select_file    => q{vcfparser_outfile_count} },
+        sv_vcfparser  => { sv_vcfparser_select_file => q{sv_vcfparser_outfile_count} },
+        vcfparser_ar  => { vcfparser_select_file    => q{vcfparser_outfile_count} },
+        fusion_report => { fusion_select_file       => q{fusion_outfile_count} },
     );
 
 ## Determine if to expect select outfile for vcfparser and sv_vcfparser

diff --git a/lib/MIP/Cli/Mip/Analyse/Rd_rna.pm b/lib/MIP/Cli/Mip/Analyse/Rd_rna.pm
@@ -211,6 +211,24 @@ sub _build_usage {
         )
     );
 
+    option(
+        q{fusion_select_file} => (
+            cmd_tags      => [q{Format: tsv; HGNC Symbol required in file}],
+            documentation => q{Select file with list of genes to analyse separately},
+            is            => q{rw},
+            isa           => Str,
+        )
+    );
+
+    option(
+        q{merge_fusion_reports} => (
+            cmd_tags      => [q{Analysis recipe switch}],
+            documentation => q{Merge sample fusion reports to case},
+            is            => q{rw},
+            isa           => enum( [ 0, 1, 2 ] ),
+        )
+    );
+
     option(
         q{fusion_use_sample_id_as_display_name} => (
             cmd_tags      => [q{Default: 0}],

diff --git a/lib/MIP/Language/Perl.pm b/lib/MIP/Language/Perl.pm
@@ -354,6 +354,7 @@ sub perl_nae_oneliners {
         q{get_fastq_header_v1.8_interleaved} => \&_get_fastq_header_v1_8_interleaved,
         get_fastq_read_length                => \&_get_fastq_read_length,
         get_gene_panel_header                => \&_get_gene_panel_info,
+        get_gene_panel_hgnc_symbols          => \&_get_gene_panel_hgnc_symbols,
         get_rrna_transcripts                 => \&_get_rrna_transcripts,
         get_select_contigs_by_col            => \&_get_select_contigs_by_col,
         get_vcf_header_id_line               => \&_get_vcf_header_id_line,
@@ -763,6 +764,23 @@ sub _get_gene_panel_info {
     return $gene_panel_info_regexp;
 }
 
+sub _get_gene_panel_hgnc_symbols {
+
+## Function : Return HGNC symbols from gene panel
+## Returns  : $gene_panel_hgnc_symbols
+## Arguments:
+
+    my ($arg_href) = @_;
+
+    # If line starts with gene panel comment
+    my $gene_panel_hgnc_symbols = q?'next if (/ \A [#] /xms ); ?;
+
+    # Append ":". Skip rest if it's a comment
+    $gene_panel_hgnc_symbols .= q?print $F[4]'?;
+
+    return $gene_panel_hgnc_symbols;
+}
+
 sub _reformat_sacct_headers {
 
 ## Function : Write individual job line - skip line containing (.batch or .bat+) in the first column

diff --git a/lib/MIP/Parameter.pm b/lib/MIP/Parameter.pm
@@ -945,6 +945,13 @@ sub set_custom_default_to_active_parameter {
                 parameter_href        => $parameter_href,
             },
         },
+        fusion_select_file => {
+            method   => \&set_default_vcfparser_select_file,
+            arg_href => {
+                active_parameter_href => $active_parameter_href,
+                parameter_name        => $parameter_name,
+            },
+        },
         infile_dirs => {
             method   => \&set_default_infile_dirs,
             arg_href => {

diff --git a/lib/MIP/Program/Gnu/Software/Gnu_grep.pm b/lib/MIP/Program/Gnu/Software/Gnu_grep.pm
@@ -15,7 +15,7 @@ use autodie qw{ :all };
 use Readonly;
 
 ## MIPs lib/
-use MIP::Constants qw{ $SPACE };
+use MIP::Constants qw{ $EQUALS $SPACE };
 use MIP::Unix::Standard_streams qw{ unix_standard_streams };
 use MIP::Unix::Write_to_file qw{ unix_write_to_file };
 
@@ -113,12 +113,12 @@ sub gnu_grep {
     }
     if ($filter_file_path) {
 
-        push @commands, q{--file=} . $filter_file_path;
+        push @commands, q{--file} . $EQUALS . $filter_file_path;
     }
 
     if ($pattern) {
 
-        push @commands, $pattern;
+        push @commands, q{--regexp} . $EQUALS . $pattern;
     }
 
     ## Infile