From 1ce218bfc0e5213d47965721ae40e7614462ef9b Mon Sep 17 00:00:00 2001
From: Frank Austin Nothaft <fnothaft@alumni.stanford.edu>
Date: Tue, 24 Oct 2017 00:27:22 -0700
Subject: [PATCH] [AVOCADO-274] Add flag to emit all genotyped variants,
 regardless of quality.

Resolves #274.
---
 .../avocado/cli/BiallelicGenotyper.scala      |  7 ++++++-
 .../avocado/util/HardFilterGenotypes.scala    | 19 ++++++++++++++-----
 2 files changed, 20 insertions(+), 6 deletions(-)

diff --git a/avocado-cli/src/main/scala/org/bdgenomics/avocado/cli/BiallelicGenotyper.scala b/avocado-cli/src/main/scala/org/bdgenomics/avocado/cli/BiallelicGenotyper.scala
index 5ac75a77..a47eb688 100644
--- a/avocado-cli/src/main/scala/org/bdgenomics/avocado/cli/BiallelicGenotyper.scala
+++ b/avocado-cli/src/main/scala/org/bdgenomics/avocado/cli/BiallelicGenotyper.scala
@@ -186,6 +186,10 @@ class BiallelicGenotyperArgs extends Args4jBase with ADAMSaveAnyArgs with Parque
     name = "-score_all_sites",
     usage = "If provided, scores all sites, even non-variant sites. Emits a gVCF styled output.")
   var scoreAllSites = false
+  @Args4jOption(required = false,
+    name = "-emit_all_genotypes",
+    usage = "Emits genotypes for all sites that were discovered.")
+  var emitAllGenotypes = false
 
   // required by HardFilterGenotypesArgs
   var maxSnpPhredStrandBias: Float = -1.0f
@@ -254,7 +258,8 @@ class BiallelicGenotyper(
 
     // hard filter the genotypes
     val filteredGenotypes = HardFilterGenotypes(genotypes, args,
-      filterRefGenotypes = !args.scoreAllSites)
+      filterRefGenotypes = !args.scoreAllSites,
+      emitAllGenotypes = (args.scoreAllSites || args.emitAllGenotypes))
 
     // save the variant calls
     filteredGenotypes.saveAsParquet(args)
diff --git a/avocado-core/src/main/scala/org/bdgenomics/avocado/util/HardFilterGenotypes.scala b/avocado-core/src/main/scala/org/bdgenomics/avocado/util/HardFilterGenotypes.scala
index 6a7dd0aa..890ae867 100644
--- a/avocado-core/src/main/scala/org/bdgenomics/avocado/util/HardFilterGenotypes.scala
+++ b/avocado-core/src/main/scala/org/bdgenomics/avocado/util/HardFilterGenotypes.scala
@@ -171,11 +171,13 @@ private[avocado] object HardFilterGenotypes extends Serializable {
    * @param grdd GenotypeRDD to filter.
    * @param args The hard filter configuration to apply.
    * @param filterRefGenotypes If true, discards homozygous ref calls.
+   * @param emitAllGenotypes If true, emits all genotypes.
    * @return A new GenotypeRDD of hard filtered genotypes.
    */
   def apply(grdd: GenotypeRDD,
             args: HardFilterGenotypesArgs,
-            filterRefGenotypes: Boolean = true): GenotypeRDD = {
+            filterRefGenotypes: Boolean = true,
+            emitAllGenotypes: Boolean = true): GenotypeRDD = {
 
     // make snp and indel filters
     val snpFilters = buildSnpHardFilters(args)
@@ -246,7 +248,8 @@ private[avocado] object HardFilterGenotypes extends Serializable {
         minQuality,
         snpFilters,
         indelFilters,
-        filterRefGenotypes))
+        filterRefGenotypes,
+        emitAllGenotypes = emitAllGenotypes))
     }).addHeaderLines(filterHeaders)
   }
 
@@ -624,6 +627,7 @@ private[avocado] object HardFilterGenotypes extends Serializable {
    * @param snpFilters Collection of filters to apply to emitted SNPs.
    * @param indelFilters Collection of filters to apply to emitted INDELs.
    * @param filterRefGenotypes If true, discards hom-ref calls.
+   * @param emitAllGenotypes If true, emits all genotypes.
    * @return If genotype is high enough quality to emit, a hard filtered
    *   genotype.
    */
@@ -632,11 +636,16 @@ private[avocado] object HardFilterGenotypes extends Serializable {
     minQuality: Int,
     snpFilters: Iterable[Genotype => Option[String]],
     indelFilters: Iterable[Genotype => Option[String]],
-    filterRefGenotypes: Boolean): Option[Genotype] = {
+    filterRefGenotypes: Boolean,
+    emitAllGenotypes: Boolean = false): Option[Genotype] = {
 
     // first, apply emission filters
-    val optGenotype = Some(genotype)
-      .filter(emitGenotypeFilter(_, minQuality, filterRefGenotypes))
+    val optGenotype = if (emitAllGenotypes) {
+      Some(genotype)
+    } else {
+      Some(genotype)
+        .filter(emitGenotypeFilter(_, minQuality, filterRefGenotypes))
+    }
 
     // then, check whether we are a snp or indel and apply hard filters
     if (genotype.getVariant.getAlternateAllele != null) {