public IEnumerable <OmimItem> GetItems()
{
var mimToGeneSymbol = GetMimNumberToGeneSymbol();
foreach (var entry in GetEntryItems())
{
int mimNumber = entry.mimNumber;
string description = entry.textSectionList?[0].textSection.textSectionContent.RemoveLinks().RemoveFormatControl();
string geneName = entry.geneMap?.geneName;
var phenotypes = entry.geneMap?.phenotypeMapList?.Select(x => OmimUtilities.GetPhenotype(x, _jsonSchema.GetSubSchema("phenotypes")))
.ToList() ?? new List <OmimItem.Phenotype>();
yield return(new OmimItem(mimToGeneSymbol[mimNumber.ToString()], geneName, description, mimNumber, phenotypes, _jsonSchema));
}
}
private IEnumerable <OmimItem> GetOmimItems(EntryRoot entryRoot, IDictionary <int, string> mimToGeneSymbol, IDictionary <int, string> phenotypeDescriptions)
{
foreach (var entry in entryRoot.omim.entryList)
{
var item = entry.entry;
var mimNumber = item.mimNumber;
//skip if not a supported gene symbol
if (!mimToGeneSymbol.TryGetValue(mimNumber, out var geneSymbol))
{
continue;
}
string description = OmimUtilities.ExtractAndProcessItemDescription(item);
string geneName = item.geneMap?.geneName;
var phenotypes = item.geneMap?.phenotypeMapList?.Select(x => OmimUtilities.GetPhenotype(x, phenotypeDescriptions, _jsonSchema.GetSubSchema("phenotypes")))
.ToList() ?? new List <OmimItem.Phenotype>();
yield return(new OmimItem(geneSymbol, geneName, description, mimNumber, phenotypes, _jsonSchema));
}
}
private static ExitCodes ProgramExecution()
{
var omimSchema = OmimSchema.Get();
var omimParser = new OmimParser(_mimToGeneFile, _omimJsonFile, omimSchema);
var version = omimParser.GetVersion();
string outFileName = $"{version.Name}_{version.Version}";
using (var nsaStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.GeneFileSuffix)))
using (var ngaWriter = new NgaWriter(nsaStream, version, SaCommon.OmimTag, SaCommon.SchemaVersion, true))
using (var saJsonSchemaStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.GeneFileSuffix + SaCommon.JsonSchemaSuffix)))
using (var schemaWriter = new StreamWriter(saJsonSchemaStream))
{
var omimItems = omimParser.GetItems();
var geneToItems = OmimUtilities.GetGeneToOmimEntriesAndSchema(omimItems);
ngaWriter.Write(geneToItems);
schemaWriter.Write(omimSchema);
}
return(ExitCodes.Success);
}
private static IDictionary <int, string> GetPhenotypeDescriptions(EntryRoot entryRoot)
{
IDictionary <int, string> phenotypeToDescription = new Dictionary <int, string>();
foreach (var entry in entryRoot.omim.entryList)
{
var item = entry.entry;
// gene only item
if (item.prefix == '*')
{
continue;
}
var description = OmimUtilities.ExtractAndProcessItemDescription(item);
if (string.IsNullOrEmpty(description))
{
continue;
}
phenotypeToDescription[item.mimNumber] = description;
}
return(phenotypeToDescription);
}
private IEnumerable <OmimItem> GetOmimItemsFromMinNumbers(IDictionary <string, string> mimToGeneSymbol, ZipArchive zipArchive)
{
var i = 0;
var queryIndex = 1;
var mimNumbers = mimToGeneSymbol.Select(x => x.Key).ToArray();
while (i < mimNumbers.Length)
{
int endMimNumberIndex = Math.Min(i + EntryQueryLimit - 1, mimNumbers.Length - 1);
string mimNumberString = GetMimNumbersString(mimNumbers, i, endMimNumberIndex);
string queryUrl = GetApiQueryUrl(OmimApiUrl, EntryHandler, ("mimNumber", mimNumberString), ("include", "text:description"), ("include", "externalLinks"), ("include", "geneMap"), ("format", ReturnDataFormat));
using (StreamWriter writer = zipArchive == null ? null : new StreamWriter(zipArchive.CreateEntry($"EntryQueryResponses/response{queryIndex}.txt").Open()))
using (var response = _httpClient.GetAsync(queryUrl).Result)
{
string responseContent = response.Content.ReadAsStringAsync().Result;
writer?.Write(responseContent);
var entryResponse = JsonConvert.DeserializeObject <EntryRoot>(responseContent);
var entryIndex = 0;
foreach (var entry in entryResponse.omim.entryList)
{
int mimNumber = entry.entry.mimNumber;
string description = OmimUtilities.RemoveLinksInText(entry.entry.textSectionList?[0].textSection.textSectionContent);
string geneName = entry.entry.geneMap?.geneName;
var phenotypes = entry.entry.geneMap?.phenotypeMapList?.Select(x => OmimUtilities.GetPhenotype(x, _jsonSchema.GetSubSchema("phenotypes"))).ToList() ??
new List <OmimItem.Phenotype>();
yield return(new OmimItem(mimToGeneSymbol[mimNumber.ToString()], geneName, description, mimNumber, phenotypes, _jsonSchema));
entryIndex++;
}
}
i = endMimNumberIndex + 1;
queryIndex++;
}
}
private static ExitCodes ProgramExecution()
{
_apiKey = GetEnvironmentVariable(OmimApiKeyEnvironmentVariableName);
if (_apiKey == null)
{
throw new InvalidDataException("Please set the OMIM API key as the environment variable \"OmimApiKey\".");
}
var version = OmimVersion.GetVersion();
string outFileName = $"{version.Name}_{version.Version}";
string dumpFilePath = Path.Combine(_outputDirectory, OmimDumpFileBaseName + version.Version + OmimDumpFileSuffix);
var(entrezGeneIdToSymbol, ensemblGeneIdToSymbol) = OmimUtilities.ParseUniversalGeneArchive(_inputReferencePath, _universalGeneArchivePath);
var geneSymbolUpdater = new GeneSymbolUpdater(entrezGeneIdToSymbol, ensemblGeneIdToSymbol);
var omimSchema = OmimSchema.Get();
using (var omimParser = new OmimParser(geneSymbolUpdater, omimSchema, _apiKey, dumpFilePath))
using (var nsaStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.NgaFileSuffix)))
using (var ngaWriter = new NgaWriter(nsaStream, version, SaCommon.OmimTag, SaCommon.SchemaVersion, true))
using (var saJsonSchemaStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.NgaFileSuffix + SaCommon.JsonSchemaSuffix)))
using (var schemaWriter = new StreamWriter(saJsonSchemaStream))
{
var omimItems = omimParser.GetItems();
var geneToItems = OmimUtilities.GetGeneToOmimEntriesAndSchema(omimItems);
ngaWriter.Write(geneToItems);
schemaWriter.Write(omimSchema);
}
geneSymbolUpdater.DisplayStatistics();
using (var writer = new StreamWriter(FileUtilities.GetCreateStream("UpdatedGeneSymbols.txt")))
{
geneSymbolUpdater.WriteUpdatedGeneSymbols(writer);
}
return(ExitCodes.Success);
}