fix: nucleotide symbol equals with dot

From GenSpectrum/cov-spectrum-website#938
GenSpectrum · Mar 5, 2024 · 6ad623e · 6ad623e
1 parent 585f646
commit 6ad623e
Show file tree

Hide file tree

Showing 6 changed files with 112 additions and 13 deletions.
diff --git a/src/silo/common/aa_symbols.cpp b/src/silo/common/aa_symbols.cpp
@@ -63,6 +63,8 @@ char AminoAcid::symbolToChar(AminoAcid::Symbol symbol) {
 
 std::optional<AminoAcid::Symbol> AminoAcid::charToSymbol(char character) {
    switch (character) {
+      case '.':
+         return std::nullopt;
       case '-':
          return AminoAcid::Symbol::GAP;
       case 'A':
@@ -114,6 +116,7 @@ std::optional<AminoAcid::Symbol> AminoAcid::charToSymbol(char character) {
       case '*':
          return AminoAcid::Symbol::STOP;
       default:
+         // TODO(#342): Revisit charToSymbol, so that illegal characters are not the same as '.'.
          return std::nullopt;
    }
 }

diff --git a/src/silo/common/aa_symbols.test.cpp b/src/silo/common/aa_symbols.test.cpp
@@ -8,6 +8,7 @@ TEST(AminoAcidSymbol, enumShouldHaveSameLengthAsArrayOfSymbols) {
 
 TEST(AminoAcidSymbol, conversionFromCharacter) {
    EXPECT_EQ(silo::AminoAcid::charToSymbol('-'), silo::AminoAcid::Symbol::GAP);
+   EXPECT_EQ(silo::AminoAcid::charToSymbol('.'), std::nullopt);
    EXPECT_EQ(silo::AminoAcid::charToSymbol('A'), silo::AminoAcid::Symbol::A);
    EXPECT_EQ(silo::AminoAcid::charToSymbol('N'), silo::AminoAcid::Symbol::N);
    EXPECT_EQ(silo::AminoAcid::charToSymbol('J'), std::nullopt);

diff --git a/src/silo/common/nucleotide_symbols.cpp b/src/silo/common/nucleotide_symbols.cpp
@@ -46,6 +46,7 @@ char Nucleotide::symbolToChar(Nucleotide::Symbol symbol) {
 std::optional<Nucleotide::Symbol> Nucleotide::charToSymbol(char character) {
    switch (character) {
       case '.':
+         return std::nullopt;
       case '-':
          return Symbol::GAP;
       case 'A':
@@ -80,6 +81,7 @@ std::optional<Nucleotide::Symbol> Nucleotide::charToSymbol(char character) {
       case 'N':
          return Symbol::N;
       default:
+         // TODO(#342): Revisit charToSymbol, so that illegal characters are not the same as '.'.
          return std::nullopt;
    }
 }

diff --git a/src/silo/common/nucleotide_symbols.test.cpp b/src/silo/common/nucleotide_symbols.test.cpp
@@ -7,7 +7,7 @@ TEST(NucleotideSymbol, enumShouldHaveSameLengthAsArrayOfSymbols) {
 }
 
 TEST(NucleotideSymbol, conversionFromCharacter) {
-   EXPECT_EQ(silo::Nucleotide::charToSymbol('.'), silo::Nucleotide::Symbol::GAP);
+   EXPECT_EQ(silo::Nucleotide::charToSymbol('.'), std::nullopt);
    EXPECT_EQ(silo::Nucleotide::charToSymbol('-'), silo::Nucleotide::Symbol::GAP);
    EXPECT_EQ(silo::Nucleotide::charToSymbol('A'), silo::Nucleotide::Symbol::A);
    EXPECT_EQ(silo::Nucleotide::charToSymbol('N'), silo::Nucleotide::Symbol::N);

diff --git a/src/silo/test/amino_acid_symbol_equals.test.cpp b/src/silo/test/amino_acid_symbol_equals.test.cpp
@@ -19,31 +19,52 @@ nlohmann::json createDataWithAminoAcidSequence(const std::string& aminoAcidSeque
    };
 }
 const nlohmann::json DATA_WITH_D = createDataWithAminoAcidSequence("D*");
-const nlohmann::json DATA_WITH_M = createDataWithAminoAcidSequence("M*");
+const nlohmann::json DATA_SAME_AS_REFERENCE = createDataWithAminoAcidSequence("M*");
 const nlohmann::json DATA_WITH_B = createDataWithAminoAcidSequence("B*");
 
 const auto DATABASE_CONFIG =
-   DatabaseConfig{"segmenet1", {"dummy name", {{"primaryKey", ValueType::STRING}}, "primaryKey"}};
+   DatabaseConfig{"segment1", {"dummy name", {{"primaryKey", ValueType::STRING}}, "primaryKey"}};
 
 const auto REFERENCE_GENOMES = ReferenceGenomes{
    {{"segment1", "A"}},
-   {{"gene1", "M*"}},
+   {{GENE, "M*"}},
 };
 
 const QueryTestData TEST_DATA{
-   {DATA_WITH_D, DATA_WITH_M, DATA_WITH_B},
+   {DATA_WITH_D, DATA_SAME_AS_REFERENCE, DATA_SAME_AS_REFERENCE, DATA_WITH_B},
    DATABASE_CONFIG,
    REFERENCE_GENOMES
 };
 
-const nlohmann::json QUERY = {
-   {"action", {{"type", "Aggregated"}}},
-   {"filterExpression",
-    {{"type", "AminoAcidEquals"}, {"position", 1}, {"symbol", "D"}, {"sequenceName", GENE}}}
-};
+nlohmann::json createAminoAcidSymbolEqualsQuery(
+   const std::string& symbol,
+   int position,
+   const std::string& gene
+) {
+   return {
+      {"action", {{"type", "Aggregated"}}},
+      {"filterExpression",
+       {{"type", "AminoAcidEquals"},
+        {"position", position},
+        {"symbol", symbol},
+        {"sequenceName", gene}}}
+   };
+}
 
-const nlohmann::json EXPECTED = {{{"count", 1}}};
+const QueryTestScenario AMINO_ACID_EQUALS_D = {
+   "aminoAcidEqualsD",
+   createAminoAcidSymbolEqualsQuery("D", 1, GENE),
+   {{{"count", 1}}}
+};
 
-const QueryTestScenario AMINO_ACID_EQUALS_D = {"aminoAcidEqualsD", QUERY, EXPECTED};
+const QueryTestScenario AMINO_ACID_EQUALS_WITH_DOT_RETURNS_AS_IF_REFERENCE = {
+   "aminoAcidEqualsM",
+   createAminoAcidSymbolEqualsQuery(".", 1, GENE),
+   {{{"count", 2}}}
+};
 
-QUERY_TEST(AminoAcidSymbolEquals, TEST_DATA, ::testing::Values(AMINO_ACID_EQUALS_D));
+QUERY_TEST(
+   AminoAcidSymbolEquals,
+   TEST_DATA,
+   ::testing::Values(AMINO_ACID_EQUALS_D, AMINO_ACID_EQUALS_WITH_DOT_RETURNS_AS_IF_REFERENCE)
+);
diff --git a/src/silo/test/nucleotide_symbol_equals.test.cpp b/src/silo/test/nucleotide_symbol_equals.test.cpp
@@ -0,0 +1,72 @@
+#include <nlohmann/json.hpp>
+
+#include "silo/test/query_fixture.test.h"
+
+using silo::ReferenceGenomes;
+using silo::config::DatabaseConfig;
+using silo::config::ValueType;
+using silo::test::QueryTestData;
+using silo::test::QueryTestScenario;
+
+#include <boost/uuid/uuid_generators.hpp>
+#include <boost/uuid/uuid_io.hpp>
+
+using boost::uuids::random_generator;
+
+nlohmann::json createDataWithNucleotideSequence(const std::string& nucleotideSequence) {
+   random_generator generator;
+   const auto request_id = generator();
+
+   return {
+      {"metadata", {{"primaryKey", "id_" + to_string(request_id)}}},
+      {"alignedNucleotideSequences", {{"segment1", nucleotideSequence}}},
+      {"unalignedNucleotideSequences", {{"segment1", nullptr}}},
+      {"alignedAminoAcidSequences", {{"gene1", nullptr}}}
+   };
+}
+const nlohmann::json DATA_SAME_AS_REFERENCE = createDataWithNucleotideSequence("ATGCN");
+const nlohmann::json DATA_WITH_ALL_N = createDataWithNucleotideSequence("NNNNN");
+const nlohmann::json DATA_WITH_ALL_MUTATED = createDataWithNucleotideSequence("CATTT");
+
+const auto DATABASE_CONFIG =
+   DatabaseConfig{"segment1", {"dummy name", {{"primaryKey", ValueType::STRING}}, "primaryKey"}};
+
+const auto REFERENCE_GENOMES = ReferenceGenomes{
+   {{"segment1", "ATGCN"}},
+   {{"gene1", "M*"}},
+};
+
+const QueryTestData TEST_DATA{
+   {DATA_SAME_AS_REFERENCE, DATA_SAME_AS_REFERENCE, DATA_WITH_ALL_N, DATA_WITH_ALL_MUTATED},
+   DATABASE_CONFIG,
+   REFERENCE_GENOMES
+};
+
+nlohmann::json createNucleotideSymbolEqualsQuery(const std::string& symbol, int position) {
+   return {
+      {"action", {{"type", "Aggregated"}}},
+      {"filterExpression",
+       {{"type", "NucleotideEquals"},
+        {"position", position},
+        {"symbol", symbol},
+        {"sequenceName", "segment1"}}}
+   };
+}
+
+const QueryTestScenario NUCLEOTIDE_EQUALS_WITH_SYMBOL = {
+   "nucleotideEqualsWithSymbol",
+   createNucleotideSymbolEqualsQuery("C", 1),
+   {{{"count", 1}}}
+};
+
+const QueryTestScenario NUCLEOTIDE_EQUALS_WITH_DOT_RETURNS_REFERENCE = {
+   "nucleotideEqualsWithDot",
+   createNucleotideSymbolEqualsQuery(".", 1),
+   {{{"count", 2}}}
+};
+
+QUERY_TEST(
+   NucleotideSymbolEquals,
+   TEST_DATA,
+   ::testing::Values(NUCLEOTIDE_EQUALS_WITH_SYMBOL, NUCLEOTIDE_EQUALS_WITH_DOT_RETURNS_REFERENCE)
+);