feat: added amino acid insertion search, added many test cases and fi…

…xed various bugs
GenSpectrum · Aug 22, 2023 · d1e4b2b · d1e4b2b
1 parent 08ae05b
commit d1e4b2b
Show file tree

Hide file tree

Showing 30 changed files with 373 additions and 239 deletions.
diff --git a/endToEndTests/test/queries/DetailsOrderBy.json b/endToEndTests/test/queries/DetailsOrderBy.json
@@ -3,9 +3,7 @@
   "query": {
     "action": {
       "type": "Details",
-      "orderByFields": [
-        "gisaid_epi_isl"
-      ]
+      "orderByFields": ["gisaid_epi_isl"]
     },
     "filterExpression": {
       "type": "And",

diff --git a/endToEndTests/test/queries/DetailsOrderByLimit.json b/endToEndTests/test/queries/DetailsOrderByLimit.json
@@ -3,9 +3,7 @@
   "query": {
     "action": {
       "type": "Details",
-      "orderByFields": [
-        "gisaid_epi_isl"
-      ],
+      "orderByFields": ["gisaid_epi_isl"],
       "offset": 9,
       "limit": 2
     },

diff --git a/endToEndTests/test/queries/LimitLargerThanTable.json b/endToEndTests/test/queries/LimitLargerThanTable.json
@@ -3,9 +3,7 @@
   "query": {
     "action": {
       "type": "Details",
-      "orderByFields": [
-        "gisaid_epi_isl"
-      ],
+      "orderByFields": ["gisaid_epi_isl"],
       "limit": 1231241
     },
     "filterExpression": {

diff --git a/endToEndTests/test/queries/Offset0.json b/endToEndTests/test/queries/Offset0.json
@@ -3,9 +3,7 @@
   "query": {
     "action": {
       "type": "Details",
-      "orderByFields": [
-        "gisaid_epi_isl"
-      ],
+      "orderByFields": ["gisaid_epi_isl"],
       "offset": 0
     },
     "filterExpression": {

diff --git a/endToEndTests/test/queries/OffsetLimitOverlap.json b/endToEndTests/test/queries/OffsetLimitOverlap.json
@@ -3,9 +3,7 @@
   "query": {
     "action": {
       "type": "Details",
-      "orderByFields": [
-        "gisaid_epi_isl"
-      ],
+      "orderByFields": ["gisaid_epi_isl"],
       "offset": 90,
       "limit": 90
     },

diff --git a/endToEndTests/test/queries/aaInsertionsAction.json b/endToEndTests/test/queries/aaInsertionsAction.json
@@ -4,10 +4,7 @@
     "action": {
       "type": "AminoAcidInsertions",
       "column": "aaInsertions",
-      "orderByFields": [
-        "insertions",
-        "position"
-      ]
+      "orderByFields": ["insertions", "position"]
     },
     "filterExpression": {
       "type": "True"
@@ -17,31 +14,31 @@
     {
       "count": 5,
       "insertions": "EPE",
-      "position": 215,
+      "position": 214,
       "sequenceName": "S"
     },
     {
       "count": 1,
       "insertions": "F",
-      "position": 3603,
+      "position": 3602,
       "sequenceName": "ORF1a"
     },
     {
       "count": 1,
       "insertions": "IV",
-      "position": 211,
+      "position": 210,
       "sequenceName": "S"
     },
     {
       "count": 1,
       "insertions": "SGE",
-      "position": 248,
+      "position": 247,
       "sequenceName": "S"
     },
     {
       "count": 1,
       "insertions": "T",
-      "position": 144,
+      "position": 143,
       "sequenceName": "S"
     }
   ]

diff --git a/endToEndTests/test/queries/aaInsertionsActionAndFilter.json b/endToEndTests/test/queries/aaInsertionsActionAndFilter.json
@@ -4,47 +4,21 @@
     "action": {
       "type": "AminoAcidInsertions",
       "column": "aaInsertions",
-      "orderByFields": [
-        "insertions",
-        "position"
-      ]
+      "orderByFields": ["insertions", "position"]
     },
     "filterExpression": {
-      "type": "InsertionContains",
+      "type": "AminoAcidInsertionContains",
+      "sequenceName": "S",
       "value": "E.*",
-      "column": "aaInsertion",
-      "position": 215
+      "column": "aaInsertions",
+      "position": 214
     }
   },
   "expectedQueryResult": [
     {
       "count": 5,
       "insertions": "EPE",
-      "position": 215,
-      "sequenceName": "S"
-    },
-    {
-      "count": 1,
-      "insertions": "F",
-      "position": 3603,
-      "sequenceName": "ORF1a"
-    },
-    {
-      "count": 1,
-      "insertions": "IV",
-      "position": 211,
-      "sequenceName": "S"
-    },
-    {
-      "count": 1,
-      "insertions": "SGE",
-      "position": 248,
-      "sequenceName": "S"
-    },
-    {
-      "count": 1,
-      "insertions": "T",
-      "position": 144,
+      "position": 214,
       "sequenceName": "S"
     }
   ]

diff --git a/endToEndTests/test/queries/aaInsertionsActionOneSequence.json b/endToEndTests/test/queries/aaInsertionsActionOneSequence.json
@@ -0,0 +1,40 @@
+{
+  "testCaseName": "amino acid insertions action only one sequence",
+  "query": {
+    "action": {
+      "type": "AminoAcidInsertions",
+      "column": "aaInsertions",
+      "orderByFields": ["insertions", "position"],
+      "sequenceName": "S"
+    },
+    "filterExpression": {
+      "type": "True"
+    }
+  },
+  "expectedQueryResult": [
+    {
+      "count": 5,
+      "insertions": "EPE",
+      "position": 214,
+      "sequenceName": "S"
+    },
+    {
+      "count": 1,
+      "insertions": "IV",
+      "position": 210,
+      "sequenceName": "S"
+    },
+    {
+      "count": 1,
+      "insertions": "SGE",
+      "position": 247,
+      "sequenceName": "S"
+    },
+    {
+      "count": 1,
+      "insertions": "T",
+      "position": 143,
+      "sequenceName": "S"
+    }
+  ]
+}
diff --git a/endToEndTests/test/queries/aaInsertionsContains.json b/endToEndTests/test/queries/aaInsertionsContains.json
@@ -0,0 +1,39 @@
+{
+  "testCaseName": "amino acid insertions contains filter",
+  "query": {
+    "action": {
+      "type": "Details",
+      "fields": ["aaInsertions", "gisaid_epi_isl"],
+      "orderByFields": ["gisaid_epi_isl"]
+    },
+    "filterExpression": {
+      "type": "AminoAcidInsertionContains",
+      "sequenceName": "S",
+      "value": "E.*E",
+      "column": "aaInsertions",
+      "position": 214
+    }
+  },
+  "expectedQueryResult": [
+    {
+      "aaInsertions": "S:214:EPE",
+      "gisaid_epi_isl": "EPI_ISL_1003629"
+    },
+    {
+      "aaInsertions": "S:214:EPE",
+      "gisaid_epi_isl": "EPI_ISL_1408408"
+    },
+    {
+      "aaInsertions": "S:214:EPE",
+      "gisaid_epi_isl": "EPI_ISL_1597932"
+    },
+    {
+      "aaInsertions": "S:214:EPE",
+      "gisaid_epi_isl": "EPI_ISL_1750868"
+    },
+    {
+      "aaInsertions": "S:214:EPE",
+      "gisaid_epi_isl": "EPI_ISL_581968"
+    }
+  ]
+}
diff --git a/endToEndTests/test/queries/insertionsAction.json b/endToEndTests/test/queries/insertionsAction.json
@@ -4,10 +4,7 @@
     "action": {
       "type": "Insertions",
       "column": "insertions",
-      "orderByFields": [
-        "insertions",
-        "position"
-      ]
+      "orderByFields": ["insertions", "position"]
     },
     "filterExpression": {
       "type": "True"
@@ -17,26 +14,26 @@
     {
       "count": 1,
       "insertions": "CAGAA",
-      "position": 22205,
-      "sequenceName": ""
+      "position": 22204,
+      "sequenceName": "main"
     },
     {
       "count": 17,
       "insertions": "CCC",
-      "position": 25702,
-      "sequenceName": ""
+      "position": 25701,
+      "sequenceName": "main"
     },
     {
       "count": 1,
       "insertions": "GCTGGT",
-      "position": 22340,
-      "sequenceName": ""
+      "position": 22339,
+      "sequenceName": "main"
     },
     {
       "count": 1,
       "insertions": "TAT",
-      "position": 5960,
-      "sequenceName": ""
+      "position": 5959,
+      "sequenceName": "main"
     }
   ]
 }
diff --git a/endToEndTests/test/queries/insertionsActionAndFilter.json b/endToEndTests/test/queries/insertionsActionAndFilter.json
@@ -16,8 +16,8 @@
     {
       "count": 1,
       "insertions": "GCTGGT",
-      "position": 22340,
-      "sequenceName": ""
+      "position": 22339,
+      "sequenceName": "main"
     }
   ]
 }