variant_builder.h

Go to the documentation of this file.

#ifndef gamgee__variant_builder__guard
#define gamgee__variant_builder__guard

#include "variant.h"
#include "variant_builder_shared_region.h"
#include "variant_builder_individual_region.h"
#include "variant_builder_multi_sample_vector.h"
#include "genotype.h"

#include <vector>
#include <string>

namespace gamgee {

template<class FIELD_TYPE>
class VariantBuilderCoreField {
 public:
  VariantBuilderCoreField() : m_field_value{}, m_is_set{false} {}

  void set(const FIELD_TYPE value) { m_field_value = value; m_is_set = true; }
  FIELD_TYPE field_value() const { return m_field_value; }
  bool is_set() const { return m_is_set; }
  void clear() { m_is_set = false; }

 private:
  FIELD_TYPE m_field_value;
  bool m_is_set;
};


class VariantBuilder {
 public:

  explicit VariantBuilder(const VariantHeader& header);

  // VariantBuilders are moveable but not copyable, and use default destruction
  VariantBuilder(VariantBuilder&& other) = default;
  VariantBuilder& operator=(VariantBuilder&& other) = default;
  VariantBuilder(const VariantBuilder& other) = delete;
  VariantBuilder& operator=(const VariantBuilder& other) = delete;
  ~VariantBuilder() = default;

  VariantHeader header() const { return VariantHeader{m_header}; }

  VariantBuilder& set_enable_validation(const bool enable_validation);


  /******************************************************************************
   *
   * Functions for setting core site-level fields
   *
   ******************************************************************************/

  VariantBuilder& set_chromosome(const uint32_t chromosome);

  VariantBuilder& set_chromosome(const std::string& chromosome);

  VariantBuilder& set_alignment_start(const uint32_t alignment_start);

  VariantBuilder& set_alignment_stop(const uint32_t alignment_stop);

  VariantBuilder& set_qual(const float qual);

  VariantBuilder& set_id(const std::string& id);

  VariantBuilder& set_ref_allele(const std::string& ref_allele);

  VariantBuilder& set_alt_allele(const std::string& alt_allele);

  VariantBuilder& set_alt_alleles(const std::vector<std::string>& alt_alleles);

  VariantBuilder& set_filters(const std::vector<std::string>& filters);

  VariantBuilder& set_filters(const std::vector<int32_t>& filters);


  /******************************************************************************
   *
   * Functions for removing core site-level fields
   * (Note that we don't have removal functions for the required fields)
   *
   ******************************************************************************/

  VariantBuilder& remove_alignment_stop();

  VariantBuilder& remove_qual();

  VariantBuilder& remove_id();

  VariantBuilder& remove_alt_alleles();

  VariantBuilder& remove_filters();


  /******************************************************************************
   *
   * Functions for setting shared/INFO fields
   *
   ******************************************************************************/

  VariantBuilder& set_integer_shared_field(const std::string& tag, const int32_t value);

  VariantBuilder& set_integer_shared_field(const std::string& tag, const std::vector<int32_t>& values);

  VariantBuilder& set_integer_shared_field(const uint32_t index, const int32_t value);

  VariantBuilder& set_integer_shared_field(const uint32_t index, const std::vector<int32_t>& values);

  VariantBuilder& set_float_shared_field(const std::string& tag, const float value);

  VariantBuilder& set_float_shared_field(const std::string& tag, const std::vector<float>& values);

  VariantBuilder& set_float_shared_field(const uint32_t index, const float value);

  VariantBuilder& set_float_shared_field(const uint32_t index, const std::vector<float>& values);

  VariantBuilder& set_string_shared_field(const std::string& tag, const std::string& value);

  VariantBuilder& set_string_shared_field(const uint32_t index, const std::string& value);

  VariantBuilder& set_boolean_shared_field(const std::string& tag);

  VariantBuilder& set_boolean_shared_field(const uint32_t index);


  /******************************************************************************
   *
   * Functions for removing shared/INFO fields
   *
   ******************************************************************************/

  VariantBuilder& remove_shared_field(const std::string& tag);

  VariantBuilder& remove_shared_field(const uint32_t field_index);

  VariantBuilder& remove_shared_fields(const std::vector<std::string>& tags);

  VariantBuilder& remove_shared_fields(const std::vector<uint32_t>& field_indices);


  /******************************************************************************
   *
   * Functions for setting individual/FORMAT fields in bulk
   * (ie., setting all values for all samples at once)
   *
   ******************************************************************************/

  VariantBuilder& set_genotypes(const VariantBuilderMultiSampleVector<int32_t>& genotypes_for_all_samples);

  VariantBuilder& set_genotypes(VariantBuilderMultiSampleVector<int32_t>&& genotypes_for_all_samples);

  VariantBuilder& set_genotypes(const std::vector<std::vector<int32_t>>& genotypes_for_all_samples);

  VariantBuilder& set_genotypes(std::vector<std::vector<int32_t>>&& genotypes_for_all_samples);

  VariantBuilder& set_integer_individual_field(const std::string& tag, const VariantBuilderMultiSampleVector<int32_t>& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const std::string& tag, VariantBuilderMultiSampleVector<int32_t>&& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const std::string& tag, const std::vector<std::vector<int32_t>>& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const std::string& tag, std::vector<std::vector<int32_t>>&& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const uint32_t field_index, const VariantBuilderMultiSampleVector<int32_t>& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const uint32_t field_index, VariantBuilderMultiSampleVector<int32_t>&& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const uint32_t field_index, const std::vector<std::vector<int32_t>>& values_for_all_samples);

  VariantBuilder& set_integer_individual_field(const uint32_t field_index, std::vector<std::vector<int32_t>>&& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const std::string& tag, const VariantBuilderMultiSampleVector<float>& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const std::string& tag, VariantBuilderMultiSampleVector<float>&& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const std::string& tag, const std::vector<std::vector<float>>& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const std::string& tag, std::vector<std::vector<float>>&& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const uint32_t field_index, const VariantBuilderMultiSampleVector<float>& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const uint32_t field_index, VariantBuilderMultiSampleVector<float>&& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const uint32_t field_index, const std::vector<std::vector<float>>& values_for_all_samples);

  VariantBuilder& set_float_individual_field(const uint32_t field_index, std::vector<std::vector<float>>&& values_for_all_samples);

  VariantBuilder& set_string_individual_field(const std::string& tag, const std::vector<std::string>& values_for_all_samples);

  VariantBuilder& set_string_individual_field(const std::string& tag, std::vector<std::string>&& values_for_all_samples);

  VariantBuilder& set_string_individual_field(const uint32_t field_index, const std::vector<std::string>& values_for_all_samples);

  VariantBuilder& set_string_individual_field(const uint32_t field_index, std::vector<std::string>&& values_for_all_samples);


  /******************************************************************************
   *
   * Functions for setting individual/FORMAT fields by sample
   *
   ******************************************************************************/

  VariantBuilder& set_genotype(const std::string& sample, const std::vector<int32_t>& genotype);

  VariantBuilder& set_genotype(const std::string& sample, std::vector<int32_t>&& genotype);

  VariantBuilder& set_genotype(const uint32_t sample_index, const std::vector<int32_t>& genotype);

  VariantBuilder& set_genotype(const uint32_t sample_index, std::vector<int32_t>&& genotype);

  VariantBuilder& set_integer_individual_field(const std::string& tag, const std::string& sample, const int32_t value);

  VariantBuilder& set_integer_individual_field(const std::string& tag, const std::string& sample, const std::vector<int32_t>& values);

  VariantBuilder& set_integer_individual_field(const uint32_t field_index, const uint32_t sample_index, const int32_t value);

  VariantBuilder& set_integer_individual_field(const uint32_t field_index, const uint32_t sample_index, const std::vector<int32_t>& values);

  VariantBuilder& set_float_individual_field(const std::string& tag, const std::string& sample, const float value);

  VariantBuilder& set_float_individual_field(const std::string& tag, const std::string& sample, const std::vector<float>& values);

  VariantBuilder& set_float_individual_field(const uint32_t field_index, const uint32_t sample_index, const float value);

  VariantBuilder& set_float_individual_field(const uint32_t field_index, const uint32_t sample_index, const std::vector<float>& values);

  VariantBuilder& set_string_individual_field(const std::string& tag, const std::string& sample, const std::string& value);

  VariantBuilder& set_string_individual_field(const uint32_t field_index, const uint32_t sample_index, const std::string& value);


  /******************************************************************************
   *
   * Functions for removing individual/FORMAT fields
   *
   ******************************************************************************/

  VariantBuilder& remove_individual_field(const std::string& tag);

  VariantBuilder& remove_individual_field(const uint32_t field_index);

  VariantBuilder& remove_individual_fields(const std::vector<std::string>& tags);

  VariantBuilder& remove_individual_fields(const std::vector<uint32_t>& field_indices);

  /******************************************************************************
   *
   * Data preparation functions for working with individual fields
   *
   ******************************************************************************/

  VariantBuilderMultiSampleVector<int32_t> get_genotype_multi_sample_vector(const uint32_t num_samples, const uint32_t max_values_per_sample) const;

  VariantBuilderMultiSampleVector<int32_t> get_integer_multi_sample_vector(const uint32_t num_samples, const uint32_t max_values_per_sample) const;

  VariantBuilderMultiSampleVector<float> get_float_multi_sample_vector(const uint32_t num_samples, const uint32_t max_values_per_sample) const;


  /******************************************************************************
   *
   * Builder operations: build() and clear()
   *
   ******************************************************************************/

  Variant build() const;

  VariantBuilder& clear();


  /******************************************************************************
   *
   * Private functionality
   *
   ******************************************************************************/

 private:
  VariantHeader m_header;
  VariantBuilderCoreField<int32_t> m_contig;
  VariantBuilderCoreField<int32_t> m_start_pos;
  VariantBuilderCoreField<int32_t> m_stop_pos;
  VariantBuilderCoreField<float> m_qual;
  VariantBuilderSharedRegion m_shared_region;
  VariantBuilderIndividualRegion m_individual_region;
  bool m_enable_validation;

  void build_from_scratch(const std::shared_ptr<bcf1_t>& new_variant_body) const;
  void post_build_validation(const std::shared_ptr<bcf1_t>& new_variant_body) const;
};

}

#endif  /* gamgee__variant_builder__guard */