Merge pull request #2032 from Clinical-Genomics/me_filter

Me filter
Clinical-Genomics · Apr 25, 2023 · 5096eff · 5096eff
2 parents 7455c5c + 3849464
commit 5096eff
Show file tree

Hide file tree

Showing 8 changed files with 701 additions and 5 deletions.
diff --git a/definitions/rd_dna_initiation_map.yaml b/definitions/rd_dna_initiation_map.yaml
@@ -55,6 +55,7 @@ CHAIN_ALL:
       - me_merge_vcfs
       - me_annotate
       - me_varianteffectpredictor
+      - me_filter
   - CHAIN_MAIN:
     # PARALLEL chains, which inherit from MAIN in initiation, but are merged back to CHAIN_MAIN after execution
     - PARALLEL:

diff --git a/definitions/rd_dna_parameters.yaml b/definitions/rd_dna_parameters.yaml
@@ -227,6 +227,7 @@ recipe_core_number:
     manta: 36
     markduplicates: 13
     me_annotate: 2
+    me_filter: 2
     me_merge_bam: 5
     me_merge_vcfs: 2
     me_varianteffectpredictor: 4
@@ -362,6 +363,7 @@ recipe_time:
     manta: 30
     markduplicates: 20
     me_annotate: 2
+    me_filter: 1
     me_merge_bam: 5
     me_merge_vcfs: 2
     me_varianteffectpredictor: 3
@@ -1366,6 +1368,25 @@ me_varianteffectpredictor:
     - tabix
     - vep
   type: recipe
+me_filter:
+  analysis_mode: case
+  associated_recipe:
+    - mip
+  data_type: SCALAR
+  default: 1
+  file_tag: _filter
+  outfile_suffix: ".vcf.gz"
+  program_executables:
+    - bcftools
+    - mip
+    - tabix
+  type: recipe
+me_filter_frequency_threshold:
+  associated_recipe:
+    - me_filter
+  data_type: SCALAR
+  default: 0.1
+  type: recipe_argument
 ## GATK CollectReadCounts
 gatk_collectreadcounts:
   analysis_mode: sample

diff --git a/lib/MIP/Cli/Mip/Analyse/Rd_dna.pm b/lib/MIP/Cli/Mip/Analyse/Rd_dna.pm
@@ -1246,6 +1246,25 @@ q{Default: hgvs, symbol, numbers, sift, polyphen, humdiv, domains, protein, ccds
         )
     );
 
+    option(
+        q{me_filter} => (
+            cmd_tags      => [q{Analysis recipe switch}],
+            documentation => q{Filter mobile elements},
+            is            => q{rw},
+            isa           => enum( [ 0, 1, 2 ] ),
+        )
+    );
+
+    option(
+        q{me_filter_frequency_threshold} => (
+            cmd_tags      => [q{Default: 0.1}],
+            documentation =>
+              q{Threshold frequency for variants to be filtered out, set to 0 to disable},
+            is  => q{rw},
+            isa => Num,
+        )
+    );
+
     option(
         q{gatk_haplotypecaller} => (
             cmd_tags      => [q{Analysis recipe switch}],