use seperate list of so terms for sv

Clinical-Genomics · Nov 25, 2022 · d2fc3bf · d2fc3bf
1 parent 0148fb8
commit d2fc3bf
Show file tree

Hide file tree

Showing 7 changed files with 214 additions and 26 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -17,6 +17,7 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 - Turn on Stringtie and gffcompare by default
 - Run varg on research vcf
 - Increase max for coverage calculation to 500x
+- Separate list of ranked SO terms for structural variants to ensure that the right SO term gets picked as the most severe for SVs
 
 ### Tools
 

diff --git a/lib/MIP/Cli/Mip/Vcfparser.pm b/lib/MIP/Cli/Mip/Vcfparser.pm
@@ -28,8 +28,7 @@ use MIP::Main::Vcfparser qw{ mip_vcfparser };
 
 command_short_description(q{MIP vcfparser command});
 
-command_long_description(
-    q{Entry point for splitting VCF into clinical and research variants});
+command_long_description(q{Entry point for splitting VCF into clinical and research variants});
 
 command_usage(q{vcfparser [options] infile.vcf [OPTIONS] > outfile.vcf});
 
@@ -60,6 +59,7 @@ sub run {
     my $padding              = $arg_href->{padding};
     my $per_gene             = $arg_href->{per_gene};
     my $pli_values_file_path = $arg_href->{pli_values_file};
+    my $variant_type         = $arg_href->{variant_type} // q{snv};
     my $write_software_tag   = $arg_href->{write_software_tag};
     my $log_file             = $arg_href->{log_file};
 
@@ -100,6 +100,7 @@ sub run {
             select_feature_file                  => $select_feature_file,
             select_feature_matching_column       => $select_feature_matching_column,
             select_outfile_path                  => $select_outfile,
+            variant_type                         => $variant_type,
             write_software_tag                   => $write_software_tag,
         }
     );
@@ -223,6 +224,14 @@ sub _build_usage {
         )
     );
 
+    option(
+        q{variant_type} => (
+            documentation => q{Variant type to parse; snv(default)/sv},
+            is            => q{rw},
+            isa           => enum( [qw{ snv sv}] ),
+        )
+    );
+
     option(
         q{write_software_tag} => (
             default       => 1,

diff --git a/lib/MIP/Constants.pm b/lib/MIP/Constants.pm
@@ -58,6 +58,7 @@ BEGIN {
       $SEMICOLON
       $SINGLE_QUOTE
       %SO_CONSEQUENCE_SEVERITY
+      %SO_CONSEQUENCE_SEVERITY_SV
       $SPACE
       $TAB
       $TEST_MODE
@@ -246,6 +247,165 @@ Readonly our %SO_CONSEQUENCE_SEVERITY => (
     },
 );
 
+Readonly our %SO_CONSEQUENCE_SEVERITY_SV => (
+    transcript_ablation => {
+        rank                      => 1,
+        genetic_region_annotation => q{exonic},
+    },
+    splice_acceptor_variant => {
+        rank                      => 2,
+        genetic_region_annotation => q{splicing},
+    },
+    splice_donor_variant => {
+        rank                      => 3,
+        genetic_region_annotation => q{splicing},
+    },
+    stop_gained => {
+        rank                      => 4,
+        genetic_region_annotation => q{exonic},
+    },
+    frameshift_variant => {
+        rank                      => 5,
+        genetic_region_annotation => q{exonic},
+    },
+    stop_lost => {
+        rank                      => 6,
+        genetic_region_annotation => q{exonic},
+    },
+    start_lost => {
+        rank                      => 7,
+        genetic_region_annotation => q{exonic},
+    },
+    coding_sequence_variant => {
+        rank                      => 8,
+        genetic_region_annotation => q{exonic},
+    },
+    transcript_amplification => {
+        rank                      => 9,
+        genetic_region_annotation => q{exonic},
+    },
+    inframe_insertion => {
+        rank                      => 10,
+        genetic_region_annotation => q{exonic},
+    },
+    inframe_deletion => {
+        rank                      => 11,
+        genetic_region_annotation => q{exonic},
+    },
+    missense_variant => {
+        rank                      => 12,
+        genetic_region_annotation => q{exonic},
+    },
+    protein_altering_variant => {
+        rank                      => 13,
+        genetic_region_annotation => q{exonic},
+    },
+    splice_region_variant => {
+        rank                      => 14,
+        genetic_region_annotation => q{splicing},
+    },
+    splice_donor_5th_base_variant => {
+        rank                      => 15,
+        genetic_region_annotation => q{splicing},
+    },
+    splice_donor_region_variant => {
+        rank                      => 16,
+        genetic_region_annotation => q{splicing},
+    },
+    splice_polypyrimidine_tract_variant => {
+        rank                      => 17,
+        genetic_region_annotation => q{splicing},
+    },
+    incomplete_terminal_codon_variant => {
+        rank                      => 18,
+        genetic_region_annotation => q{exonic},
+    },
+    non_coding_transcript_exon_variant => {
+        rank                      => 19,
+        genetic_region_annotation => q{ncRNA_exonic},
+    },
+    synonymous_variant => {
+        rank                      => 20,
+        genetic_region_annotation => q{exonic},
+    },
+    start_retained_variant => {
+        rank                      => 21,
+        genetic_region_annotation => q{exonic},
+    },
+    stop_retained_variant => {
+        rank                      => 22,
+        genetic_region_annotation => q{exonic},
+    },
+    mature_miRNA_variant => {
+        rank                      => 23,
+        genetic_region_annotation => q{ncRNA_exonic},
+    },
+    q{5_prime_UTR_variant} => {
+        rank                      => 24,
+        genetic_region_annotation => q{5UTR},
+    },
+    q{3_prime_UTR_variant} => {
+        rank                      => 25,
+        genetic_region_annotation => q{3UTR},
+    },
+    intron_variant => {
+        rank                      => 26,
+        genetic_region_annotation => q{intronic},
+    },
+    NMD_transcript_variant => {
+        rank                      => 27,
+        genetic_region_annotation => q{ncRNA},
+    },
+    non_coding_transcript_variant => {
+        rank                      => 28,
+        genetic_region_annotation => q{ncRNA},
+    },
+    upstream_gene_variant => {
+        rank                      => 29,
+        genetic_region_annotation => q{upstream},
+    },
+    downstream_gene_variant => {
+        rank                      => 30,
+        genetic_region_annotation => q{downstream},
+    },
+    TFBS_ablation => {
+        rank                      => 31,
+        genetic_region_annotation => q{TFBS},
+    },
+    TFBS_amplification => {
+        rank                      => 32,
+        genetic_region_annotation => q{TFBS},
+    },
+    TF_binding_site_variant => {
+        rank                      => 33,
+        genetic_region_annotation => q{TFBS},
+    },
+    regulatory_region_ablation => {
+        rank                      => 34,
+        genetic_region_annotation => q{regulatory_region},
+    },
+    regulatory_region_amplification => {
+        rank                      => 35,
+        genetic_region_annotation => q{regulatory_region},
+    },
+    feature_elongation => {
+        rank                      => 36,
+        genetic_region_annotation => q{genomic_feature},
+    },
+    regulatory_region_variant => {
+        rank                      => 37,
+        genetic_region_annotation => q{regulatory_region},
+    },
+    feature_truncation => {
+        rank                      => 38,
+        genetic_region_annotation => q{genomic_feature},
+    },
+    intergenic_variant => {
+        rank                      => 39,
+        genetic_region_annotation => q{intergenic},
+    },
+);
+
 ## Contigs
 Readonly our %PRIMARY_CONTIG => (
     38 => {

diff --git a/lib/MIP/Main/Vcfparser.pm b/lib/MIP/Main/Vcfparser.pm
@@ -15,7 +15,8 @@ use autodie qw{ :all };
 use Readonly;
 
 ## MIPs lib/
-use MIP::Constants qw{ $COLON $LOG_NAME $NEWLINE %SO_CONSEQUENCE_SEVERITY $TAB };
+use MIP::Constants
+  qw{ $COLON $LOG_NAME $NEWLINE %SO_CONSEQUENCE_SEVERITY %SO_CONSEQUENCE_SEVERITY_SV $TAB };
 use MIP::File::Format::Feature_file qw{ read_feature_file };
 use MIP::File::Format::Pli qw{ load_pli_file };
 use MIP::Log::MIP_log4perl qw{ retrieve_log };
@@ -46,6 +47,7 @@ sub mip_vcfparser {
 ##          : $select_feature_file                   => Select feature file
 ##          : $select_feature_matching_column        => Select feature matching column
 ##          : $select_outfile_path                   => Select file path
+##          : $variant_type                          => Type of variants to parse
 ##          : $vcf_in_fh                             => VCF in filehandle
 ##          : $write_software_tag                    => Write software tag to vcf header switch
 
@@ -56,6 +58,7 @@ sub mip_vcfparser {
     my $pli_values_file_path;
     my $range_feature_annotation_columns_ref;
     my $range_feature_file;
+    my $variant_type;
     my $select_feature_annotation_columns_ref;
     my $select_feature_matching_column;
     my $select_outfile_path;
@@ -114,9 +117,14 @@ sub mip_vcfparser {
             store       => \$select_feature_matching_column,
             strict_type => 1,
         },
-        select_outfile_path => { store   => \$select_outfile_path, strict_type => 1, },
-        vcf_in_fh           => { defined => 1, required => 1, store => \$vcf_in_fh, },
-        write_software_tag  => {
+        select_outfile_path => { store => \$select_outfile_path, strict_type => 1, },
+        variant_type        => {
+            allow       => [qw{ snv sv }],
+            store       => \$variant_type,
+            strict_type => 1,
+        },
+        vcf_in_fh          => { defined => 1, required => 1, store => \$vcf_in_fh, },
+        write_software_tag => {
             allow       => [ 0, 1 ],
             default     => 1,
             store       => \$write_software_tag,
@@ -127,7 +135,8 @@ sub mip_vcfparser {
     check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
 
     ## Constants
-    my %consequence_severity = %SO_CONSEQUENCE_SEVERITY;
+    my %consequence_severity =
+      ( $variant_type eq q{sv} ) ? %SO_CONSEQUENCE_SEVERITY_SV : %SO_CONSEQUENCE_SEVERITY;
 
     my ( %meta_data, %range_data, %tree, %pli_score );
 

diff --git a/lib/MIP/Program/Mip.pm b/lib/MIP/Program/Mip.pm
@@ -305,6 +305,7 @@ sub mip_vcfparser {
 ##          : $stderrfile_path                       => Stderrfile path
 ##          : $stderrfile_path_append                => Append stderr info to file path
 ##          : $stdoutfile_path                       => Stdoutfile path
+##          : $variant_type                          => Type of variants to parse <snv/sv>
 
     my ($arg_href) = @_;
 
@@ -322,6 +323,7 @@ sub mip_vcfparser {
     my $stderrfile_path;
     my $stderrfile_path_append;
     my $stdoutfile_path;
+    my $variant_type;
 
     ## Default(s)
     my $padding;
@@ -393,6 +395,11 @@ sub mip_vcfparser {
             store       => \$stdoutfile_path,
             strict_type => 1,
         },
+        variant_type => {
+            allow       => [ undef, qw{snv sv} ],
+            store       => \$variant_type,
+            strict_type => 1,
+        },
     };
 
     check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
@@ -465,6 +472,11 @@ sub mip_vcfparser {
         push @commands, q{--select_outfile} . $SPACE . $select_outfile;
     }
 
+    if ($variant_type) {
+
+        push @commands, q{--variant_type} . $SPACE . $variant_type;
+    }
+
     push @commands,
       unix_standard_streams(
         {

diff --git a/lib/MIP/Recipes/Analysis/Mip_vcfparser.pm b/lib/MIP/Recipes/Analysis/Mip_vcfparser.pm
@@ -1073,6 +1073,7 @@ sub analysis_mip_vcfparser_sv_wes {
             select_feature_file_path              => $select_file,
             select_feature_matching_column        => $select_file_matching_column,
             select_outfile                        => $select_outfile,
+            variant_type                          => q{sv},
         }
     );
     say {$filehandle} $NEWLINE;
@@ -1411,6 +1412,7 @@ sub analysis_mip_vcfparser_sv_wgs {
                 select_feature_file_path       => $select_file,
                 select_feature_matching_column => $select_file_matching_column,
                 select_outfile                 => $select_outfile,
+                variant_type                   => q{sv},
             }
         );