GL-18 Funcotation filter for homvar & compound hets on two autosomal recessive genes #5843
Conversation
Codecov Report
@@ Coverage Diff @@
## master #5843 +/- ##
===============================================
- Coverage 87.064% 87.058% -0.005%
- Complexity 32123 32206 +83
===============================================
Files 1977 1984 +7
Lines 147336 147583 +247
Branches 16199 16227 +28
===============================================
+ Hits 128276 128483 +207
- Misses 13154 13177 +23
- Partials 5906 5923 +17
|
|
@als364 Can I request that you title your pull requests more descriptively? Ie., include the tool involved in the PR title, and a concise one-line description of the improvements/fixes. This would help when writing GATK release notes, as we use the PR titles as the basis for initial release notes. Thanks! |
|
Yes, can do. |
als364
changed the title
| @@ -213,10 +231,56 @@ private VariantContext applyFilters(final VariantContext variant, final Set<Stri | |||
| return variantContextBuilder.make(); | |||
| } | |||
|
|
|||
| private void buildArHetByGene(VariantContext variant) { | |||
There was a problem hiding this comment.
Should this be in its own class?. Having AR specific code in the FilterFuncotations class seems like it breaks encapsulation.
There was a problem hiding this comment.
Yeah, it could be. There's not currently a 'good' place for it as this can't be a FuncotationFilter.
There was a problem hiding this comment.
Yeah - it may be worth designing another class hierarchy for filters that require two passes. That way the next time someone has to make a filter that requires work on the first pass it'll be easier.
Maybe as a subclass of FuncotationFilter?
There was a problem hiding this comment.
This looks pretty good.
I had a couple of comments - one about a refactor (similar to what @tlangs suggested), and a few others about some coding conventions we typically use for GATK tools.
src/main/java/org/broadinstitute/hellbender/tools/funcotator/FilterFuncotations.java
Outdated
Show resolved
Hide resolved
| @@ -213,10 +231,56 @@ private VariantContext applyFilters(final VariantContext variant, final Set<Stri | |||
| return variantContextBuilder.make(); | |||
| } | |||
|
|
|||
| private void buildArHetByGene(VariantContext variant) { | |||
There was a problem hiding this comment.
Yeah - it may be worth designing another class hierarchy for filters that require two passes. That way the next time someone has to make a filter that requires work on the first pass it'll be easier.
Maybe as a subclass of FuncotationFilter?
src/main/java/org/broadinstitute/hellbender/tools/funcotator/FilterFuncotations.java
Outdated
Show resolved
Hide resolved
...main/java/org/broadinstitute/hellbender/tools/funcotator/filtrationRules/ArHomvarFilter.java
Outdated
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/tools/funcotator/FilterFuncotations.java
Outdated
Show resolved
Hide resolved
...n/java/org/broadinstitute/hellbender/tools/funcotator/filtrationRules/FuncotationFilter.java
Outdated
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/tools/funcotator/FilterFuncotations.java
Show resolved
Hide resolved
| import java.util.stream.Stream; | ||
|
|
||
| public class FilterFuncotationsUtils { | ||
| public static Stream<Set<Map.Entry<String, String>>> getTranscriptFuncotations(final FuncotationMap funcotationMap) { |
There was a problem hiding this comment.
Could you add Javadoc for this method?
| } | ||
|
|
||
| private boolean arHetvarRule(Set<Map.Entry<String, String>> funcotations, VariantContext variant) { | ||
| return arCompoundHetVariants.stream().anyMatch(hetVariant -> variantContextsMatch(hetVariant, variant)); |
There was a problem hiding this comment.
Might want to add in a warning/error here (or in the parent class) if the firstPassApply and/or afterFirstPass was never called.
| && v1.getStart() == v2.getStart() | ||
| && v1.getEnd() == v2.getEnd() | ||
| && v1.getReference() == v2.getReference() | ||
| && v1.getReference() == v2.getReference() |
There was a problem hiding this comment.
Looks like this line is a copy/paste error - was there something else you wanted to test between the variant contexts?
This PR adds a new filter for
FilterFuncotations. For two autosomal recessive genes,MUTYHandATP7B, homozygous variants and compound heterozygous variants will be tagged and added to the output vcf.