vcf4.4 initial changes

EBIvariation · Mar 12, 2024 · 313458b · 313458b
1 parent b1d57e5
commit 313458b
Show file tree

Hide file tree

Showing 327 changed files with 39,627 additions and 45 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -216,6 +216,7 @@ set (MOD_VCF_SOURCES
   inc/vcf/validator_detail_v41.hpp
   inc/vcf/validator_detail_v42.hpp
   inc/vcf/validator_detail_v43.hpp
+  inc/vcf/validator_detail_v44.hpp
   inc/vcf/validator.hpp
 
   src/fasta/faidx.cpp
@@ -293,6 +294,7 @@ endif ()
 set (V41_TESTS test/vcf/parser_v41_test.cpp)
 set (V42_TESTS test/vcf/parser_v42_test.cpp)
 set (V43_TESTS test/vcf/parser_v43_test.cpp)
+set (V44_TESTS test/vcf/parser_v44_test.cpp)
 set (ALL_TESTS
   test/assembly_report/assembly_report_test.cpp
   test/fasta/faidx_test.cpp
@@ -309,6 +311,7 @@ set (ALL_TESTS
   test/vcf/parser_v41_test.cpp
   test/vcf/parser_v42_test.cpp
   test/vcf/parser_v43_test.cpp
+  test/vcf/parser_v44_test.cpp
   test/vcf/predefined_info_tags_test.cpp
   test/vcf/predefined_format_tags_test.cpp
   test/vcf/record_cache_test.cpp
@@ -332,6 +335,11 @@ target_link_libraries (test_validator_v43 ${LIBRARIES_TO_LINK})
 enable_testing ()
 add_test (NAME ValidatorTests_v43 COMMAND test_validator_v43)
 
+add_executable (test_validator_v44 test/main_test.cpp ${V44_TESTS})
+target_link_libraries (test_validator_v44 ${LIBRARIES_TO_LINK})
+enable_testing ()
+add_test (NAME ValidatorTests_v44 COMMAND test_validator_v44)
+
 add_executable (test_validation_suite test/main_test.cpp ${ALL_TESTS})
 target_link_libraries (test_validation_suite ${LIBRARIES_TO_LINK})
 enable_testing ()

diff --git a/appveyor.yml b/appveyor.yml
@@ -18,6 +18,7 @@ test_script:
   - cmd: bin\test_validator_v41.exe
   - cmd: bin\test_validator_v42.exe
   - cmd: bin\test_validator_v43.exe
+  - cmd: bin\test_validator_v44.exe
   - cmd: bin\test_validation_suite.exe
 
 artifacts:

diff --git a/docs/developer-guide.md b/docs/developer-guide.md
@@ -41,6 +41,10 @@ executable binaries:
 
 ![architecture image](img/VCF-validator-architecture.png)
 
+In case of dependent libraries being installed in non-standard paths,
+update LIBRARY_PATH as shown below
+  'export LIBRARY_PATH=$LIBRARY_PATH:<path to required library's libpath>'
+
 The role of Ragel and ODB in this project is explained further below.
 
 ## Code structure

diff --git a/inc/vcf/file_structure.hpp b/inc/vcf/file_structure.hpp
@@ -55,7 +55,8 @@ namespace ebi
     {
         v41,
         v42,
-        v43
+        v43,
+        v44
     };
 
     enum class RecordType
@@ -159,6 +160,53 @@ namespace ebi
             { THOUSAND_G, { FLAG, "0" } }
     };
 
+    const std::map<std::string, std::pair<std::string, std::string>> info_v44 = {
+            { AA, { STRING, "1" } },
+            { AC, { INTEGER, A } },
+            { AD, { INTEGER, R } },
+            { ADF, { INTEGER, R } },
+            { ADR, { INTEGER, R } },
+            { AF, { FLOAT, A } },
+            { AN, { INTEGER, "1" } },
+            { BKPTID, { STRING, UNKNOWN_CARDINALITY } },
+            { BQ, { FLOAT, "1" } },
+            { CICN, { INTEGER, "2" } },
+            { CICNADJ, { INTEGER, UNKNOWN_CARDINALITY } },
+            { CIEND, { INTEGER, "2" } },
+            { CIGAR, { STRING, A } },
+            { CILEN, { INTEGER, "2" } },
+            { CIPOS, { INTEGER, "2" } },
+            { CN, { INTEGER, "1" } },
+            { CNADJ, { INTEGER, UNKNOWN_CARDINALITY } },
+            { DB, { FLAG, "0" } },
+            { DBRIPID, { STRING, "1" } },
+            { DBVARID, { STRING, "1" } },
+            { DGVID, { STRING, "1" } },
+            { DP, { INTEGER, "1" } },
+            { DPADJ, { INTEGER, UNKNOWN_CARDINALITY } },
+            { END, { INTEGER, "1" } },
+            { EVENT, { STRING, "1" } },
+            { HOMLEN, { INTEGER, UNKNOWN_CARDINALITY } },
+            { HOMSEQ, { STRING, UNKNOWN_CARDINALITY } },
+            { H2, { FLAG, "0" } },
+            { H3, { FLAG, "0" } },
+            { IMPRECISE, { FLAG, "0" } },
+            { MATEID, { STRING, UNKNOWN_CARDINALITY } },
+            { MEINFO, { STRING, "4" } },
+            { METRANS, { STRING, "4" } },
+            { MQ, { MISSING_VALUE, "1" } },
+            { MQ0, { INTEGER, "1" } },
+            { NOVEL, { FLAG, "0" } },
+            { NS, { INTEGER, "1" } },
+            { PARID, { STRING, "1" } },
+            // TODO : SB metadata Type and Number is "."
+            { SOMATIC, { FLAG, "0" } },
+            { SVLEN, { INTEGER, UNKNOWN_CARDINALITY } },
+            { SVTYPE, { STRING, "1" } },
+            { VALIDATED, { FLAG, "0" } },
+            { THOUSAND_G, { FLAG, "0" } }
+    };
+
     const std::map<std::string, std::pair<std::string, std::string>> format_v41_v42 = {
             { AHAP, { INTEGER, "1" } },
             { CN, { INTEGER, "1" } },
@@ -206,6 +254,31 @@ namespace ebi
             { PS, { INTEGER, "1" } }
     };
 
+    const std::map<std::string, std::pair<std::string, std::string>> format_v44 = {
+            { AD, { INTEGER, R } },
+            { ADF, { INTEGER, R } },
+            { ADR, { INTEGER, R } },
+            { AHAP, { INTEGER, "1" } },
+            { CN, { INTEGER, "1" } },
+            { CNL, { FLOAT, G } },
+            { CNP, { FLOAT, G } },
+            { CNQ, { FLOAT, "1" } },
+            { DP, { INTEGER, "1" } },
+            { EC, { INTEGER, A } },
+            { FT, { STRING, "1" } },
+            { GL, { FLOAT, G } },
+            { GP, { FLOAT, G } },
+            { GQ, { INTEGER, "1" } },
+            { GT, { STRING, "1" } },
+            { HAP, { INTEGER, "1" } },
+            { HQ, { INTEGER, "2" } },
+            { MQ, { INTEGER, "1" } },
+            { NQ, { INTEGER, "1" } },
+            { PL, { INTEGER, G } },
+            { PQ, { INTEGER, "1" } },
+            { PS, { INTEGER, "1" } }
+    };
+
     const std::set<std::string> PREDEFINED_INFO_SVTYPES{
             DEL,
             INS,

diff --git a/inc/vcf/string_constants.hpp b/inc/vcf/string_constants.hpp
@@ -27,6 +27,7 @@ namespace ebi
     const std::string VCF_V41 = "VCFv4.1";
     const std::string VCF_V42 = "VCFv4.2";
     const std::string VCF_V43 = "VCFv4.3";
+    const std::string VCF_V44 = "VCFv4.4";
 
     // validator, debugulator and assembly_checker command line arguments
     const char STDIN[] = "stdin";

diff --git a/inc/vcf/validator.hpp b/inc/vcf/validator.hpp
@@ -172,6 +172,24 @@ namespace ebi
         void parse_buffer(char const * p, char const * pe, char const * eof);
     };
 
+    template <typename Configuration>
+    class ParserImpl_v44
+    : public ParserImpl,
+      Configuration::ParsePolicy,
+      Configuration::ErrorPolicy,
+      Configuration::OptionalPolicy
+    {
+      public:
+        using ParsePolicy = typename Configuration::ParsePolicy;
+        using ErrorPolicy = typename Configuration::ErrorPolicy;
+        using OptionalPolicy = typename Configuration::OptionalPolicy;
+
+        ParserImpl_v44(std::shared_ptr<Source> source, AdditionalChecks additionalChecks);
+
+      private:
+        void parse_buffer(char const * p, char const * pe, char const * eof);
+    };
+
     // Predefined aliases for common uses of the parser
     using QuickValidator_v41 = ParserImpl_v41<QuickValidatorCfg>;
     using FullValidator_v41 = ParserImpl_v41<FullValidatorCfg>;
@@ -185,6 +203,10 @@ namespace ebi
     using FullValidator_v43 = ParserImpl_v43<FullValidatorCfg>;
     using Reader_v43 = ParserImpl_v43<ReaderCfg>;
 
+    using QuickValidator_v44 = ParserImpl_v44<QuickValidatorCfg>;
+    using FullValidator_v44 = ParserImpl_v44<FullValidatorCfg>;
+    using Reader_v44 = ParserImpl_v44<ReaderCfg>;
+
     bool is_valid_vcf_file(std::istream &input, const std::string &sourceName,
                            ValidationLevel validationLevel,
                            std::vector<std::unique_ptr<ebi::vcf::ReportWriter>> &outputs,
@@ -210,5 +232,6 @@ namespace ebi
 #include "validator_detail_v41.hpp"
 #include "validator_detail_v42.hpp"
 #include "validator_detail_v43.hpp"
+#include "validator_detail_v44.hpp"
 
 #endif // VCF_VALIDATOR_HPP