public patient AddPatient(string mrn, string mrnSource, string firstName, string lastName, DateTime? dateOfBirth)
{
var result = (from existingPatient in entities.patients
where existingPatient.external_source == mrnSource && existingPatient.external_id == mrn
select existingPatient).FirstOrDefault();
if (result == null)
{
patient newPatient = new patient()
{
external_source = mrnSource,
external_id = mrn,
date_of_birth = dateOfBirth,
first_name = firstName,
last_name = lastName
};
entities.patients.Add(newPatient);
entities.SaveChanges();
result = newPatient;
}
return result;
}
public result_entities AddPhenotype(result_files resultFile, patient patient, attribute attribute, DateTime? resultedOn)
{
result_entities rootEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = attribute.id,
};
result_entities resultedOnEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = GetAttribute(null, null, "Resulted on", null).id,
parent = rootEntity,
value_date_time = resultedOn
};
entities.result_entities.AddRange(new[] { rootEntity, resultedOnEntity });
entities.SaveChanges();
return rootEntity;
}
public override void LoadData(string filePath)
{
var vcfParser = new VCFParser(filePath);
var header = vcfParser.Header;
var patient = new patient();
var source = sourceRepo.AddSource("VCF", "VCF file");
var file = AddResultFile(filePath, source);
// Process the file-level pragmas
result_entities rootEntity = new result_entities()
{
attribute_id = EntityRepository.GetAttribute(null, null, "Variant Call Format result", null).id,
result_file_id = file.id
};
// We pull out all of the metadata from the header (all lines) and write them as information
// lines associated with this result.
var headerEntities = new List<result_entities>();
foreach (var headerItem in header.MetaDataInInputOrder)
{
if (headerItem.Key == "individual-id")
{
var individualParts = headerItem.Value.Replace("<", "").Replace(">", "").Split(new char[] { ',' });
var individualData = individualParts.Select(x => x.Split(new char[] { '=' })).ToArray();
var mrnParts = individualData.FirstOrDefault(x => x[0] == "Dbxref")[1].Split(':');
patient = patientRepo.AddPatient(mrnParts[1], mrnParts[0],
individualData.FirstOrDefault(x => x[0] == "First_name")[1],
individualData.FirstOrDefault(x => x[0] == "Last_name")[1],
DateTime.Parse(individualData.FirstOrDefault(x => x[0] == "DOB")[1]));
}
else if (headerItem.Key == "fileDate")
{
DateTime resultDate = DateTime.ParseExact(headerItem.Value, "yyyyMMdd", CultureInfo.InvariantCulture, DateTimeStyles.None);
headerEntities.Add(CreateEntityAttribute("Resulted on", 0, file.id, rootEntity, resultDate.ToShortDateString()));
}
else if (headerItem.GetType() == typeof(VCFInfoHeaderLine))
{
var info = headerItem as VCFInfoHeaderLine;
var infoEntity = CreateEntityAttribute("INFO", 0, file.id, rootEntity, null);
headerEntities.Add(infoEntity);
headerEntities.Add(CreateEntityAttribute("ID", 0, file.id, infoEntity, info.ID));
headerEntities.Add(CreateEntityAttribute("Number", 0, file.id, infoEntity, info.CountType.ToString()));
headerEntities.Add(CreateEntityAttribute("Type", 0, file.id, infoEntity, info.Type.ToString()));
headerEntities.Add(CreateEntityAttribute("Description", 0, file.id, infoEntity, info.Description));
}
else if (headerItem.GetType() == typeof(VCFFilterHeaderLine))
{
var filter = headerItem as VCFFilterHeaderLine;
var filterEntity = CreateEntityAttribute("FILTER", 0, file.id, rootEntity, null);
headerEntities.Add(filterEntity);
foreach (var field in filter.GenericFields())
{
headerEntities.Add(CreateEntityAttribute(field.Key, 0, file.id, filterEntity, field.Value));
}
}
else if (headerItem.GetType() == typeof(VCFFormatHeaderLine))
{
var format = headerItem as VCFFormatHeaderLine;
var formatEntity = CreateEntityAttribute("FORMAT", 0, file.id, rootEntity, null);
headerEntities.Add(formatEntity);
headerEntities.Add(CreateEntityAttribute("ID", 0, file.id, formatEntity, format.ID));
headerEntities.Add(CreateEntityAttribute("Number", 0, file.id, formatEntity, format.CountType.ToString()));
headerEntities.Add(CreateEntityAttribute("Type", 0, file.id, formatEntity, format.Type.ToString()));
headerEntities.Add(CreateEntityAttribute("Description", 0, file.id, formatEntity, format.Description));
}
else
{
var headerEntity = CreateEntityAttribute(headerItem.Key, 0, file.id, rootEntity, headerItem.Value);
headerEntities.Add(headerEntity);
}
}
rootEntity.patient_id = patient.id;
headerEntities.ForEach(x => x.patient_id = patient.id);
var variantEntities = new List<result_entities>();
while (vcfParser.MoveNext())
{
var current = vcfParser.Current;
result_entities variantEntity = new result_entities()
{
attribute_id = EntityRepository.GetAttribute(null, null, "Variant Call Format variant", null).id,
result_file_id = file.id,
patient_id = patient.id,
parent = rootEntity
};
variantEntities.Add(variantEntity);
result_entities snpEntity = new result_entities()
{
attribute_id = EntityRepository.GetAttribute(current.ID, "dbSNP", null, null).id,
result_file_id = file.id,
patient_id = patient.id,
parent = variantEntity
};
variantEntities.Add(snpEntity);
SetVariantValues(current, patient.id, file.id, snpEntity, variantEntities);
variantEntities.Add(CreateEntityAttribute("Chromosome", patient.id, file.id, variantEntity, current.Chr));
variantEntities.Add(CreateEntityAttribute("Start position", patient.id, file.id, variantEntity, current.Start.ToString()));
variantEntities.Add(CreateEntityAttribute("End position", patient.id, file.id, variantEntity, current.End.ToString()));
variantEntities.Add(CreateEntityAttribute("Reference base", patient.id, file.id, variantEntity, current.Reference.BaseString));
variantEntities.Add(CreateEntityAttribute("Quality", patient.id, file.id, variantEntity, current.PhredScaledQual.ToString()));
foreach (var attr in current.Attributes)
{
variantEntities.Add(CreateEntityAttribute(string.Format("INFO:{0}", attr.Key), patient.id, file.id, variantEntity, attr.Value.ToString()));
}
if (current.FiltersMaybeNull != null)
{
foreach (var filter in current.FiltersMaybeNull)
{
variantEntities.Add(CreateEntityAttribute(string.Format("FILTER:{0}", filter), patient.id, file.id, variantEntity, string.Empty));
}
}
//foreach (var genotype in current.Genotypes)
//{
// attributeList.Add(AddVariantInformation("VCF:Genotype", genotype.ToMHGRString()));
//}
//attributeList.Add(AddVariantInformation("VCF:Quality", current.PhredScaledQual.ToString()));
//attributeList.Add(AddVariantInformation("VCF:Filter", string.Join(",", current.Filters.ToArray())));
//featureInformationList.Add(patientVariant, attributeList);
}
entityRepo.AddVCF(rootEntity, headerEntities, variantEntities);
}
public result_entities[] AddSnps(result_files resultFile, patient patient, DateTime? resultedOn, List<SnpResult> snps)
{
var results = new List<result_entities>();
var snpAlleleAttributeId = GetAttribute(null, null, "SNP allele", null).id;
var referenceBaseAttributeId = GetAttribute(null, null, "Reference base", null).id;
var resultedOnAttributeId = GetAttribute(null, null, "Resulted on", null).id;
foreach (var snp in snps)
{
var entityParts = new List<result_entities>();
result_entities rootEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = GetAttribute(snp.RSID, "dbSNP", snp.RSID, snp.RSID).id,
};
result_entities resultedOnEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = resultedOnAttributeId,
parent = rootEntity,
value_date_time = resultedOn
};
result_entities referenceBaseEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = referenceBaseAttributeId,
parent = rootEntity,
value_short_text = snp.ReferenceBase
};
result_entities allele1Entity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = snpAlleleAttributeId,
parent = rootEntity,
value_short_text = snp.Genotype[0].ToString()
};
result_entities allele2Entity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = snpAlleleAttributeId,
parent = rootEntity,
value_short_text = snp.Genotype[1].ToString()
};
entities.result_entities.AddRange(new[] { rootEntity, resultedOnEntity, referenceBaseEntity, allele1Entity, allele2Entity });
results.Add(rootEntity);
}
entities.SaveChanges();
return results.ToArray();
}
public result_entities[] AddStarVariants(patient patient, result_files resultFile, DateTime resultedOn, List<StarVariantResult> stars)
{
var results = new List<result_entities>();
foreach (var star in stars)
{
var entityParts = new List<result_entities>();
result_entities rootEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = GetAttribute(star.Gene, "HGNC", star.Gene, star.Gene).id
};
entityParts.Add(rootEntity);
result_entities resultedOnEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = GetAttribute(null, null, "Resulted on", null).id,
parent = rootEntity,
value_date_time = resultedOn
};
entityParts.Add(resultedOnEntity);
string[] splitStars = star.Result.Split(new string[] { "*" }, StringSplitOptions.RemoveEmptyEntries);
foreach (var value in splitStars)
{
result_entities alleleEntity = new result_entities()
{
patient_id = patient.id,
result_file_id = resultFile.id,
attribute_id = GetAttribute(null, null, "Star allele", null).id,
parent = rootEntity,
value_short_text = value
};
entityParts.Add(alleleEntity);
}
entities.result_entities.AddRange(entityParts);
results.Add(rootEntity);
}
entities.SaveChanges();
return results.ToArray();
}
