Merge pull request #146 from GenSpectrum/testsPosIndex

fix: test cases verifying that the positions index for mutation distribution are now 1-indexed
GenSpectrum · Jul 4, 2023 · 69fdc2b · 69fdc2b
2 parents 76e0823 + fdf972a
commit 69fdc2b
Show file tree

Hide file tree

Showing 6 changed files with 276 additions and 0 deletions.
diff --git a/endToEndTests/test/invalidQueries/sequencePos0Filter.json b/endToEndTests/test/invalidQueries/sequencePos0Filter.json
@@ -0,0 +1,17 @@
+{
+  "testCaseName": "Filtering for the position 0",
+  "query": {
+    "action": {
+      "type": "Aggregated"
+    },
+    "filterExpression": {
+      "position": 0,
+      "symbol": "-",
+      "type": "NucleotideEquals"
+    }
+  },
+  "expectedError": {
+    "error": "Bad request",
+    "message": "The field 'position' in a NucleotideEquals expression needs to be an unsigned integer greater than 0"
+  }
+}
diff --git a/endToEndTests/test/queries/sequenceEndFilter.json b/endToEndTests/test/queries/sequenceEndFilter.json
@@ -0,0 +1,18 @@
+{
+  "testCaseName": "Filtering for the last genome position",
+  "query": {
+    "action": {
+      "type": "Aggregated"
+    },
+    "filterExpression": {
+      "position": 29903,
+      "symbol": "-",
+      "type": "NucleotideEquals"
+    }
+  },
+  "expectedQueryResult": [
+    {
+      "count": 45
+    }
+  ]
+}
diff --git a/endToEndTests/test/queries/sequenceStartEndMutations.json b/endToEndTests/test/queries/sequenceStartEndMutations.json
@@ -0,0 +1,216 @@
+{
+  "testCaseName": "Getting a mutation distribution that contains the first and last genome position",
+  "query": {
+    "action": {
+      "type": "Mutations",
+      "minProportion": 1
+    },
+    "filterExpression": {
+      "children": [
+        {
+          "position": 1,
+          "symbol": "-",
+          "type": "NucleotideEquals"
+        },
+        {
+          "position": 29903,
+          "symbol": "-",
+          "type": "NucleotideEquals"
+        }
+      ],
+      "type": "And"
+    }
+  },
+  "expectedQueryResult": [
+    {
+      "count": 42,
+      "position": "A1-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "C3037T",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "C14408T",
+      "proportion": 1
+    },
+    {
+      "count": 40,
+      "position": "A23403G",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "C29870-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29871-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29872-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29873-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29874-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29875-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29876-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29877-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29878-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29879-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29880-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29881-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29882-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29883-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29884-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29885-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29886-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29887-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29888-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29889-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29890-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29891-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29892-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29893-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29894-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29895-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29896-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29897-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29898-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29899-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29900-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29901-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29902-",
+      "proportion": 1
+    },
+    {
+      "count": 42,
+      "position": "A29903-",
+      "proportion": 1
+    }
+  ]
+}
diff --git a/endToEndTests/test/queries/sequenceStartFilter.json b/endToEndTests/test/queries/sequenceStartFilter.json
@@ -0,0 +1,18 @@
+{
+  "testCaseName": "Filtering for the first genome position",
+  "query": {
+    "action": {
+      "type": "Aggregated"
+    },
+    "filterExpression": {
+      "position": 1,
+      "symbol": "-",
+      "type": "NucleotideEquals"
+    }
+  },
+  "expectedQueryResult": [
+    {
+      "count": 45
+    }
+  ]
+}
diff --git a/src/silo/query_engine/filter_expressions/nucleotide_symbol_equals.cpp b/src/silo/query_engine/filter_expressions/nucleotide_symbol_equals.cpp
@@ -28,6 +28,12 @@ std::unique_ptr<silo::query_engine::operators::Operator> NucleotideSymbolEquals:
    const silo::DatabasePartition& database_partition,
    Expression::AmbiguityMode mode
 ) const {
+   if (position >= database.reference_genome->genome_segments[0].size()) {
+      throw QueryParseException(
+         "NucleotideEquals position is out of bounds '" + std::to_string(position + 1) + "' > '" +
+         std::to_string(database.reference_genome->genome_segments[0].size()) + "'"
+      );
+   }
    NUCLEOTIDE_SYMBOL nucleotide_symbol;
    if (value == '.') {
       const char character = database.reference_genome->genome_segments[0].at(position);

diff --git a/src/silo/storage/pango_lineage_alias.cpp b/src/silo/storage/pango_lineage_alias.cpp
@@ -3,6 +3,7 @@
 #include <filesystem>
 #include <fstream>
 #include <iterator>
+#include <sstream>
 
 #include <spdlog/spdlog.h>
 #include <nlohmann/json.hpp>