Merge pull request #198 from andresfsilva/feature/check-genotype-freq…

…uency

EVA-1199 Validate genotypes and frequencies are present

The flag `--require-evidence` can be passed to the vcf validator to indicate it should check the VCF has either Genotypes or Allele Frequencies

README.md

@@ -45,6 +45,8 @@ Different types of validation reports can be written with the `-r` / `--report`
 
 Each report is written into its own file and it is named after the input file, followed by a timestamp. The default output directory is the same as the input file's if provided using `-i`, or the current directory if using the standard input; it can be changed with the `-o` / `--outdir` option.
 
+A flag for validating evidence can be passed with `--require-evidence`. This will validate whether the VCF includes either Genotypes or Allele Frequencies. This flag will be **false** by default.
+
 ### Debugulator
 
 There are some simple errors that can be automatically fixed. The most common error is the presence of duplicate variants. The needed parameters are the original VCF and the report generated by a previous run of the validator with the option `-r database`.

inc/vcf/parsing_state.hpp

@@ -30,7 +30,10 @@ namespace ebi
 {
   namespace vcf
   {
-
+    struct AdditionalChecks {
+        bool checkEvidence;
+    };
+
     struct ParsingState
     {
         size_t n_lines;
@@ -46,8 +49,9 @@ namespace ebi
         std::vector<std::unique_ptr<Error>> warnings;
 
         std::multimap<std::string, std::string> defined_metadata;
+        AdditionalChecks additionalChecks;
 
-        ParsingState(std::shared_ptr<Source> source);
+        ParsingState(std::shared_ptr<Source> source, AdditionalChecks additionalChecks);
         virtual ~ParsingState() = default;
 
         void set_version(Version version);
@@ -66,6 +70,14 @@ namespace ebi
         bool is_well_defined_meta(std::string const & meta_type, std::string const & id) const;
 
         void add_well_defined_meta(std::string const & meta_type, std::string const & id);
+
+        void validate_additional_checks();
+
+        bool genotypes_present();
+
+        bool allele_frequencies_present();
+
+        bool allele_count_present();
     };
   }
 }

inc/vcf/string_constants.hpp

@@ -23,7 +23,6 @@ namespace ebi
 {
   namespace vcf
   {
-
     // VCF versions
     const std::string VCF_V41 = "VCFv4.1";
     const std::string VCF_V42 = "VCFv4.2";
@@ -63,6 +62,7 @@ namespace ebi
     const char REPORT_OPTION[] = "report,r";
     const char OUTDIR_OPTION[] = "outdir,o";
     const char OUTPUT_OPTION[] = "output,o";
+    const char CHECK_EVIDENCE[] = "require-evidence";
 
     // fields
     const std::string ID = "ID";

inc/vcf/validator.hpp

@@ -96,7 +96,7 @@ namespace ebi
       public ParsingState
     {
       public:
-        ParserImpl(std::shared_ptr<Source> source);
+        ParserImpl(std::shared_ptr<Source> source, AdditionalChecks additionalChecks);
         virtual ~ParserImpl() = default;
 
         void parse(std::string const & text) override;
@@ -130,7 +130,7 @@ namespace ebi
         using ErrorPolicy = typename Configuration::ErrorPolicy;
         using OptionalPolicy = typename Configuration::OptionalPolicy;
 
-        ParserImpl_v41(std::shared_ptr<Source> source);
+        ParserImpl_v41(std::shared_ptr<Source> source, AdditionalChecks additionalChecks);
 
       private:
         void parse_buffer(char const * p, char const * pe, char const * eof);
@@ -148,7 +148,7 @@ namespace ebi
         using ErrorPolicy = typename Configuration::ErrorPolicy;
         using OptionalPolicy = typename Configuration::OptionalPolicy;
 
-        ParserImpl_v42(std::shared_ptr<Source> source);
+        ParserImpl_v42(std::shared_ptr<Source> source, AdditionalChecks additionalChecks);
 
       private:
         void parse_buffer(char const * p, char const * pe, char const * eof);
@@ -166,7 +166,7 @@ namespace ebi
         using ErrorPolicy = typename Configuration::ErrorPolicy;
         using OptionalPolicy = typename Configuration::OptionalPolicy;
 
-        ParserImpl_v43(std::shared_ptr<Source> source);
+        ParserImpl_v43(std::shared_ptr<Source> source, AdditionalChecks additionalChecks);
 
       private:
         void parse_buffer(char const * p, char const * pe, char const * eof);
@@ -185,15 +185,16 @@ namespace ebi
     using FullValidator_v43 = ParserImpl_v43<FullValidatorCfg>;
     using Reader_v43 = ParserImpl_v43<ReaderCfg>;
 
-    bool is_valid_vcf_file(std::istream &input,
-                           const std::string &sourceName,
+    bool is_valid_vcf_file(std::istream &input, const std::string &sourceName,
                            ValidationLevel validationLevel,
-                           std::vector<std::unique_ptr<ebi::vcf::ReportWriter>> &outputs);
+                           std::vector<std::unique_ptr<ebi::vcf::ReportWriter>> &outputs,
+                           AdditionalChecks checks);
 
     bool process_vcf_stream(std::istream &input,
-                          const std::string &sourceName,
-                          ValidationLevel validationLevel,
-                          std::vector<std::unique_ptr<ebi::vcf::ReportWriter>> &outputs);
+                            const std::string &sourceName,
+                            ValidationLevel validationLevel,
+                            std::vector<std::unique_ptr<ebi::vcf::ReportWriter>> &outputs,
+                            AdditionalChecks additionalChecks);
 
     std::string get_compression_from_extension(std::string const & source);