Use policy enum in get_called_alleles rather than bool switch

Daniel Cooke · Daniel Cooke · commit a1262e2a14d6 · 2019-03-14T17:11:33.000Z
Also fixes a bug in NormalContamination and MedianSomaticMappingQuality resulting from trimming reference alleles.
diff --git a/src/core/csr/measures/base_mismatch_count.cpp b/src/core/csr/measures/base_mismatch_count.cpp
@@ -82,7 +82,7 @@ Measure::ResultType BaseMismatchCount::do_evaluate(const VcfRecord& call, const
     result.reserve(samples.size());
     for (const auto& sample : samples) {
         std::vector<Allele> alleles; bool has_ref;
-        std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+        std::tie(alleles, has_ref) = get_called_alleles(call, sample);
         int sample_result {0};
         if (!alleles.empty()) {
             const auto sample_allele_support = compute_allele_support(alleles, assignments, sample);
diff --git a/src/core/csr/measures/denovo_contamination.cpp b/src/core/csr/measures/denovo_contamination.cpp
@@ -63,9 +63,9 @@ void sort_unique(Container& values)
 
 auto get_denovo_alleles(const VcfRecord& denovo, const Trio& trio)
 {
-    auto parent_alleles = concat(get_called_alleles(denovo, trio.mother(), true).first,
-                                 get_called_alleles(denovo, trio.father(), true).first);
-    auto child_alleles = get_called_alleles(denovo, trio.child(), true).first;
+    auto parent_alleles = concat(get_called_alleles(denovo, trio.mother()).first,
+                                 get_called_alleles(denovo, trio.father()).first);
+    auto child_alleles = get_called_alleles(denovo, trio.child()).first;
     sort_unique(parent_alleles); sort_unique(child_alleles);
     std::vector<Allele> result {};
     result.reserve(child_alleles.size());
diff --git a/src/core/csr/measures/mapping_quality_divergence.cpp b/src/core/csr/measures/mapping_quality_divergence.cpp
@@ -145,7 +145,7 @@ Measure::ResultType MappingQualityDivergence::do_evaluate(const VcfRecord& call,
         boost::optional<int> sample_result {};
         if (call.is_heterozygous(sample)) {
             std::vector<Allele> alleles; bool has_ref;
-            std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+            std::tie(alleles, has_ref) = get_called_alleles(call, sample);
             if (!alleles.empty()) {
                 const auto sample_allele_support = compute_allele_support(alleles, assignments, sample);
                 const auto mapping_qualities = extract_mapping_qualities(sample_allele_support);
diff --git a/src/core/csr/measures/median_base_quality.cpp b/src/core/csr/measures/median_base_quality.cpp
@@ -79,7 +79,7 @@ Measure::ResultType MedianBaseQuality::do_evaluate(const VcfRecord& call, const
         boost::optional<int> sample_result {};
         if (is_evaluable(call, sample)) {
             std::vector<Allele> alleles; bool has_ref;
-            std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+            std::tie(alleles, has_ref) = get_called_alleles(call, sample);
             if (has_ref) alleles.erase(std::cbegin(alleles));
             if (!alleles.empty()) {
                 const auto sample_allele_support = compute_allele_support(alleles, assignments, sample);
diff --git a/src/core/csr/measures/median_somatic_mapping_quality.cpp b/src/core/csr/measures/median_somatic_mapping_quality.cpp
@@ -35,6 +35,11 @@ std::unique_ptr<Measure> MedianSomaticMappingQuality::do_clone() const
 
 namespace {
 
+auto extract_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample)
+{
+    return get_called_alleles(call, sample, ReferencePadPolicy::trim_alt_alleles).first;
+}
+
 template <typename Container>
 void sort_unique(Container& values)
 {
@@ -47,10 +52,10 @@ auto get_somatic_alleles(const VcfRecord& somatic, const std::vector<SampleName>
 {
     std::vector<Allele> somatic_sample_alleles {}, normal_sample_alleles {};
     for (const auto& sample : somatic_samples) {
-        utils::append(get_called_alleles(somatic, sample, true).first, somatic_sample_alleles);
+        utils::append(extract_called_alleles(somatic, sample), somatic_sample_alleles);
     }
     for (const auto& sample : normal_samples) {
-        utils::append(get_called_alleles(somatic, sample, true).first, normal_sample_alleles);
+        utils::append(extract_called_alleles(somatic, sample), normal_sample_alleles);
     }
     sort_unique(somatic_sample_alleles); sort_unique(normal_sample_alleles);
     std::vector<Allele> result {};
diff --git a/src/core/csr/measures/mismatch_count.cpp b/src/core/csr/measures/mismatch_count.cpp
@@ -57,7 +57,7 @@ Measure::ResultType MismatchCount::do_evaluate(const VcfRecord& call, const Face
     result.reserve(samples.size());
     for (const auto& sample : samples) {
         std::vector<Allele> alleles; bool has_ref;
-        std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+        std::tie(alleles, has_ref) = get_called_alleles(call, sample);
         int sample_result {0};
         if (!alleles.empty()) {
             const auto sample_allele_support = compute_allele_support(alleles, assignments, sample);
diff --git a/src/core/csr/measures/normal_contamination.cpp b/src/core/csr/measures/normal_contamination.cpp
@@ -33,6 +33,11 @@ std::unique_ptr<Measure> NormalContamination::do_clone() const
 
 namespace {
 
+auto extract_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample)
+{
+    return get_called_alleles(call, sample, ReferencePadPolicy::trim_alt_alleles).first;
+}
+
 template <typename Container>
 void sort_unique(Container& values)
 {
@@ -45,10 +50,10 @@ auto get_somatic_alleles(const VcfRecord& somatic, const std::vector<SampleName>
 {
     std::vector<Allele> somatic_sample_alleles {}, normal_sample_alleles {};
     for (const auto& sample : somatic_samples) {
-        utils::append(get_called_alleles(somatic, sample, true).first, somatic_sample_alleles);
+        utils::append(extract_called_alleles(somatic, sample), somatic_sample_alleles);
     }
     for (const auto& sample : normal_samples) {
-        utils::append(get_called_alleles(somatic, sample, true).first, normal_sample_alleles);
+        utils::append(extract_called_alleles(somatic, sample), normal_sample_alleles);
     }
     sort_unique(somatic_sample_alleles); sort_unique(normal_sample_alleles);
     std::vector<Allele> result {};
diff --git a/src/core/csr/measures/read_end_bias.cpp b/src/core/csr/measures/read_end_bias.cpp
@@ -136,7 +136,7 @@ Measure::ResultType ReadEndBias::do_evaluate(const VcfRecord& call, const FacetM
     const EndDefinition end_def {end_fraction_};
     for (const auto& sample : samples) {
         std::vector<Allele> alleles; bool has_ref;
-        std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+        std::tie(alleles, has_ref) = get_called_alleles(call, sample);
         if (!alleles.empty()) {
             const auto allele_support = compute_allele_support(alleles, assignments, sample);
             result.push_back(calculate_max_tail_bias(allele_support, end_def));
diff --git a/src/core/csr/measures/read_side_bias.cpp b/src/core/csr/measures/read_side_bias.cpp
@@ -115,7 +115,7 @@ Measure::ResultType ReadSideBias::do_evaluate(const VcfRecord& call, const Facet
     result.reserve(samples.size());
     for (const auto& sample : samples) {
         std::vector<Allele> alleles; bool has_ref;
-        std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+        std::tie(alleles, has_ref) = get_called_alleles(call, sample);
         if (!alleles.empty()) {
             const auto allele_support = compute_allele_support(alleles, assignments, sample);
             auto position_bias = calculate_position_bias(allele_support);
diff --git a/src/core/csr/measures/read_tail_bias.cpp b/src/core/csr/measures/read_tail_bias.cpp
@@ -126,7 +126,7 @@ Measure::ResultType ReadTailBias::do_evaluate(const VcfRecord& call, const Facet
     const TailDefinition tail_def {tail_fraction_};
     for (const auto& sample : samples) {
         std::vector<Allele> alleles; bool has_ref;
-        std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+        std::tie(alleles, has_ref) = get_called_alleles(call, sample);
         if (!alleles.empty()) {
             const auto allele_support = compute_allele_support(alleles, assignments, sample);
             result.push_back(calculate_max_tail_bias(allele_support, tail_def));
diff --git a/src/core/csr/measures/strand_bias.cpp b/src/core/csr/measures/strand_bias.cpp
@@ -173,7 +173,7 @@ Measure::ResultType StrandBias::do_evaluate(const VcfRecord& call, const FacetMa
         boost::optional<double> sample_result {};
         if (is_evaluable(call, sample)) {
             std::vector<Allele> alleles; bool has_ref;
-            std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+            std::tie(alleles, has_ref) = get_called_alleles(call, sample);
             assert(!alleles.empty());
             const auto sample_allele_support = compute_allele_support(alleles, assignments, sample);
             const auto direction_counts = get_direction_counts(sample_allele_support, mapped_region(call));
diff --git a/src/core/csr/measures/supplementary_fraction.cpp b/src/core/csr/measures/supplementary_fraction.cpp
@@ -32,7 +32,7 @@ Measure::ResultType SupplementaryFraction::do_evaluate(const VcfRecord& call, co
     result.reserve(samples.size());
     for (const auto& sample : samples) {
         std::vector<Allele> alleles; bool has_ref;
-        std::tie(alleles, has_ref) = get_called_alleles(call, sample, true);
+        std::tie(alleles, has_ref) = get_called_alleles(call, sample);
         if (has_ref) alleles.erase(std::cbegin(alleles));
         boost::optional<double> sample_result {};
         if (!alleles.empty()) {
diff --git a/src/utils/genotype_reader.cpp b/src/utils/genotype_reader.cpp
@@ -169,7 +169,7 @@ auto extract_genotype(const VcfRecord& call, const SampleName& sample)
 } // namespace
 
 std::pair<std::vector<Allele>, bool>
-get_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample, const bool trim_padding)
+get_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample, const ReferencePadPolicy ref_pad_policy)
 {
     auto genotype = get_genotype(call, sample);
     remove_missing_alleles(genotype);
@@ -182,7 +182,7 @@ get_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample, c
     std::vector<Allele> result {};
     result.reserve(genotype.size());
     bool has_ref {false};
-    if (trim_padding) {
+    if (ref_pad_policy != ReferencePadPolicy::leave) {
         auto first_itr = std::begin(genotype);
         const auto ref_itr = std::find(first_itr, std::end(genotype), call.ref());
         if (ref_itr != std::end(genotype)) {
@@ -214,9 +214,12 @@ get_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample, c
         }
         if (has_ref) {
             auto& ref = genotype.front();
-            ref.erase(std::cbegin(ref), std::next(std::cbegin(ref), *max_ref_pad));
-            auto allele_region = expand_lhs(call_region, -*max_ref_pad);
-            result.emplace_back(std::move(allele_region), std::move(ref));
+            if (ref_pad_policy == ReferencePadPolicy::trim_all_alleles) {
+                ref.erase(std::cbegin(ref), std::next(std::cbegin(ref), *max_ref_pad));
+                result.emplace_back(expand_lhs(call_region, -*max_ref_pad), std::move(ref));
+            } else {
+                result.emplace_back(call_region, std::move(ref));
+            }
             std::rotate(std::rbegin(result), std::next(std::rbegin(result)), std::rend(result));
         }
         if (!unknwown_pad_allele_indices.empty()) {
diff --git a/src/utils/genotype_reader.hpp b/src/utils/genotype_reader.hpp
@@ -26,13 +26,17 @@ class GenomicRegion;
 
 using GenotypeMap = std::unordered_map<SampleName, MappableFlatSet<Genotype<Haplotype>>>;
 
-std::pair<std::vector<Allele>, bool>
-get_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample, const bool trim_padding = true);
+enum class ReferencePadPolicy { leave, trim_alt_alleles, trim_all_alleles };
 
-GenotypeMap extract_genotypes(const std::vector<VcfRecord>& calls,
-                              const std::vector<VcfRecord::SampleName>& samples,
-                              const ReferenceGenome& reference,
-                              boost::optional<GenomicRegion> call_region = boost::none);
+std::pair<std::vector<Allele>, bool>
+get_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample,
+                   ReferencePadPolicy ref_pad_policy = ReferencePadPolicy::trim_all_alleles);
+
+GenotypeMap
+extract_genotypes(const std::vector<VcfRecord>& calls,
+                  const std::vector<VcfRecord::SampleName>& samples,
+                  const ReferenceGenome& reference,
+                  boost::optional<GenomicRegion> call_region = boost::none);
 
 }
 

Original file line number	Diff line number	Diff line change
`@@ -35,6 +35,11 @@ std::unique_ptr<Measure> MedianSomaticMappingQuality::do_clone() const`
`35`	`35`
`36`	`36`	`namespace {`
`37`	`37`
	`38`	`+auto extract_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample)`
	`39`	`+{`
	`40`	`+ return get_called_alleles(call, sample, ReferencePadPolicy::trim_alt_alleles).first;`
	`41`	`+}`
	`42`	`+`
`38`	`43`	`template <typename Container>`
`39`	`44`	`void sort_unique(Container& values)`
`40`	`45`	`{`
`@@ -47,10 +52,10 @@ auto get_somatic_alleles(const VcfRecord& somatic, const std::vector<SampleName>`
`47`	`52`	`{`
`48`	`53`	`std::vector<Allele> somatic_sample_alleles {}, normal_sample_alleles {};`
`49`	`54`	`for (const auto& sample : somatic_samples) {`
`50`		`- utils::append(get_called_alleles(somatic, sample, true).first, somatic_sample_alleles);`
	`55`	`+ utils::append(extract_called_alleles(somatic, sample), somatic_sample_alleles);`
`51`	`56`	`}`
`52`	`57`	`for (const auto& sample : normal_samples) {`
`53`		`- utils::append(get_called_alleles(somatic, sample, true).first, normal_sample_alleles);`
	`58`	`+ utils::append(extract_called_alleles(somatic, sample), normal_sample_alleles);`
`54`	`59`	`}`
`55`	`60`	`sort_unique(somatic_sample_alleles); sort_unique(normal_sample_alleles);`
`56`	`61`	`std::vector<Allele> result {};`
Original file line number	Diff line number	Diff line change
`@@ -33,6 +33,11 @@ std::unique_ptr<Measure> NormalContamination::do_clone() const`
`33`	`33`
`34`	`34`	`namespace {`
`35`	`35`
	`36`	`+auto extract_called_alleles(const VcfRecord& call, const VcfRecord::SampleName& sample)`
	`37`	`+{`
	`38`	`+ return get_called_alleles(call, sample, ReferencePadPolicy::trim_alt_alleles).first;`
	`39`	`+}`
	`40`	`+`
`36`	`41`	`template <typename Container>`
`37`	`42`	`void sort_unique(Container& values)`
`38`	`43`	`{`
`@@ -45,10 +50,10 @@ auto get_somatic_alleles(const VcfRecord& somatic, const std::vector<SampleName>`
`45`	`50`	`{`
`46`	`51`	`std::vector<Allele> somatic_sample_alleles {}, normal_sample_alleles {};`
`47`	`52`	`for (const auto& sample : somatic_samples) {`
`48`		`- utils::append(get_called_alleles(somatic, sample, true).first, somatic_sample_alleles);`
	`53`	`+ utils::append(extract_called_alleles(somatic, sample), somatic_sample_alleles);`
`49`	`54`	`}`
`50`	`55`	`for (const auto& sample : normal_samples) {`
`51`		`- utils::append(get_called_alleles(somatic, sample, true).first, normal_sample_alleles);`
	`56`	`+ utils::append(extract_called_alleles(somatic, sample), normal_sample_alleles);`
`52`	`57`	`}`
`53`	`58`	`sort_unique(somatic_sample_alleles); sort_unique(normal_sample_alleles);`
`54`	`59`	`std::vector<Allele> result {};`