// Protocol buffers to store the extracted variants from ClinVar.

syntax = "proto3";

package annonars.clinvar_data.extracted_vars;

import "annonars/clinvar_data/clinvar_public.proto";

// Enumeration for the type of the variant.
enum VariationType {
  // unspecified variation type
  VARIATION_TYPE_UNSPECIFIED = 0;
  // Corresponds to "insertion".
  VARIATION_TYPE_INSERTION = 1;
  // Corresponds to "deletion".
  VARIATION_TYPE_DELETION = 2;
  // Corresponds to "single nucleotide variant".
  VARIATION_TYPE_SNV = 3;
  // Corresponds to "indel".
  VARIATION_TYPE_INDEL = 4;
  // Corresponds to "duplication".
  VARIATION_TYPE_DUPLICATION = 5;
  // Corresponds to "tandem duplication".
  VARIATION_TYPE_TANDEM_DUPLICATION = 6;
  // Corresponds to "structural variant".
  VARIATION_TYPE_STRUCTURAL_VARIANT = 7;
  // Corresponds to "copy number gain".
  VARIATION_TYPE_COPY_NUMBER_GAIN = 8;
  // Corresponds to "copy number loss".
  VARIATION_TYPE_COPY_NUMBER_LOSS = 9;
  // Corresponds to "protein only".
  VARIATION_TYPE_PROTEIN_ONLY = 10;
  // Corresponds to "microsatellite".
  VARIATION_TYPE_MICROSATELLITE = 11;
  // Corresponds to "inversion".
  VARIATION_TYPE_INVERSION = 12;
  // Corresponds to "other".
  VARIATION_TYPE_OTHER = 13;
}

// Accession with version.
message VersionedAccession {
  // The accession.
  string accession = 1;
  // The version.
  int32 version = 2;
}

// Protocol buffer for storing essential information of one RCV.
message ExtractedRcvRecord {
  // The accession.
  VersionedAccession accession = 1;
  // Title of RCV.
  string title = 2;
  // Classifications (thinned out).
  annonars.clinvar_data.clinvar_public.RcvAccession.RcvClassifications classifications = 3;
}

// Protocol buffer for storing essential information of one VCV.
message ExtractedVcvRecord {
  // The accession.
  VersionedAccession accession = 1;
  // List of aggregated RCVs.
  repeated ExtractedRcvRecord rcvs = 2;
  // Name of VCV.
  string name = 3;
  // The type of the variant.
  VariationType variation_type = 4;
  // Classifications (thinned out).
  annonars.clinvar_data.clinvar_public.AggregateClassificationSet classifications = 5;
  // Clinical assertions (thinned out),
  repeated annonars.clinvar_data.clinvar_public.ClinicalAssertion clinical_assertions = 8;
  // The sequence location on one reference.
  annonars.clinvar_data.clinvar_public.Location.SequenceLocation sequence_location = 6;
  // List of HGNC IDs.
  repeated string hgnc_ids = 7;
}