public void GenerateMimToGeneSymbolFile(GeneSymbolUpdater geneSymbolUpdater)
{
using StreamWriter writer = new StreamWriter(_mimToSymbolStream);
using var response = _httpClient.GetAsync(Mim2GeneUrl).Result;
using var reader = new StreamReader(response.Content.ReadAsStreamAsync().Result);
writer.WriteLine("#MIM number\tGene symbol");
string line;
while ((line = reader.ReadLine()) != null)
{
if (line.OptimizedStartsWith('#'))
{
continue;
}
var fields = line.OptimizedSplit('\t');
var geneSymbol = fields[3];
if (string.IsNullOrEmpty(geneSymbol))
{
continue;
}
var mimNumber = fields[0];
var entrezGeneId = fields[2];
var ensemblGeneId = fields[4];
var updatedGeneSymbol = geneSymbolUpdater.UpdateGeneSymbol(geneSymbol, ensemblGeneId, entrezGeneId);
if (string.IsNullOrEmpty(updatedGeneSymbol))
{
continue;
}
writer.WriteLine($"{mimNumber}\t{updatedGeneSymbol}");
}
}
public OmimParser(GeneSymbolUpdater geneSymbolUpdater, SaJsonSchema jsonSchema, string apiKey, string dumpFilePath)
{
_geneSymbolUpdater = geneSymbolUpdater;
_jsonSchema = jsonSchema;
if (dumpFilePath != null)
{
_zipFileStream = new FileStream(dumpFilePath, FileMode.Create);
_zipArchive = new ZipArchive(_zipFileStream, ZipArchiveMode.Create);
}
_httpClient = new HttpClient();
_httpClient.DefaultRequestHeaders.Add("ApiKey", apiKey);
}
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(entrezGeneIdToSymbol, ensemblGeneIdToSymbol) = GeneUtilities.ParseUniversalGeneArchive(_inputReferencePath, _universalGeneArchivePath);
var geneSymbolUpdater = new GeneSymbolUpdater(entrezGeneIdToSymbol, ensemblGeneIdToSymbol);
using (var omimQuery = new OmimQuery(_apiKey, _outputDirectory))
{
omimQuery.GenerateMimToGeneSymbolFile(geneSymbolUpdater);
omimQuery.GenerateJsonResponse();
}
OmimVersion.WriteToFile(OmimQuery.JsonResponseFile, _outputDirectory);
geneSymbolUpdater.DisplayStatistics();
return(ExitCodes.Success);
}
public IDictionary <string, string> GenerateMimToGeneSymbol(GeneSymbolUpdater geneSymbolUpdater)
{
var mimNumberToGeneSymbol = new Dictionary <string, string>();
using (StreamWriter writer = new StreamWriter(_minToSymbolStream))
using (var response = _httpClient.GetAsync(Mim2GeneUrl).Result)
using (var reader = new StreamReader(response.Content.ReadAsStreamAsync().Result))
{
writer.WriteLine("#MIM number\tGene symbol");
string line;
while ((line = reader.ReadLine()) != null)
{
if (line.OptimizedStartsWith('#'))
{
continue;
}
var fields = line.OptimizedSplit('\t');
string geneSymbol = fields[3];
if (string.IsNullOrEmpty(geneSymbol))
{
continue;
}
string mimNumber = fields[0];
string entrezGeneId = fields[2];
string ensemblGeneId = fields[4];
string updatedGeneSymbol = geneSymbolUpdater.UpdateGeneSymbol(geneSymbol, ensemblGeneId, entrezGeneId);
if (string.IsNullOrEmpty(updatedGeneSymbol))
{
continue;
}
writer.WriteLine($"{mimNumber}\t{updatedGeneSymbol}");
mimNumberToGeneSymbol[mimNumber] = updatedGeneSymbol;
}
}
return(mimNumberToGeneSymbol);
}
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);
}