public List <MutableGene> Flatten(int overlapStart = -1, int overlapEnd = -1)
{
var combinedGenes = new List <MutableGene>();
var genesById = GeneUtilities.GetGenesById(_genes, _isEnsembl);
foreach (var gene in _genes)
{
if (gene.Invalid)
{
continue;
}
var geneId = _isEnsembl
? gene.EnsemblId.ToString()
: gene.EntrezGeneId.ToString();
List <MutableGene> genesWithSameGeneId;
if (!genesById.TryGetValue(geneId, out genesWithSameGeneId))
{
throw new UserErrorException($"Unable to find similar genes for {geneId}");
}
combinedGenes.Add(GetFlattenedGene(gene, genesWithSameGeneId, overlapStart, overlapEnd));
}
if (_showOutput)
{
Console.WriteLine($" - {_description}: {combinedGenes.Count} genes.");
}
return(combinedGenes);
}
public List <MutableGene> Merge()
{
_mergedGenes.Clear();
var genesBySymbol = GeneUtilities.GetGenesBySymbol(_genes);
foreach (var gene in _genes)
{
if (gene.Invalid)
{
continue;
}
List <MutableGene> genesWithSameSymbol;
if (!genesBySymbol.TryGetValue(gene.Symbol, out genesWithSameSymbol))
{
throw new UserErrorException($"Unable to find similar genes for {gene.Symbol}");
}
MergesGenesWithSameSymbol(gene, genesWithSameSymbol);
}
Console.WriteLine($" - {_numOrphanEnsemblGenes} orphan Ensembl genes.");
Console.WriteLine($" - {_numOrphanRefSeqGenes} orphan RefSeq genes.");
Console.WriteLine($" - {_numMergedGenes} merged genes.");
return(_mergedGenes);
}
private void MergesGenesWithSameSymbol(MutableGene seedGene, List <MutableGene> genesWithSameSymbol)
{
int overlapStart, overlapEnd;
var validGenes = GetValidGenes(seedGene, genesWithSameSymbol, out overlapStart, out overlapEnd);
var ensemblGenes = GeneUtilities.GetGenesByDataSource(validGenes, TranscriptDataSource.Ensembl);
var refSeqGenes = GeneUtilities.GetGenesByDataSource(validGenes, TranscriptDataSource.RefSeq);
var ensemblFlattener = new GeneFlattener(ensemblGenes, "Ensembl", false);
var flatEnsemblGenes = ensemblFlattener.Flatten(overlapStart, overlapEnd);
var refSeqFlattener = new GeneFlattener(refSeqGenes, "RefSeq", false);
var flatRefSeqGenes = refSeqFlattener.Flatten(overlapStart, overlapEnd);
foreach (var ensemblGene in flatEnsemblGenes)
{
// add the unused Ensembl genes
string linkedEntrezId;
if (!_linkedEnsemblIds.TryGetValue(ensemblGene.EnsemblId.ToString(), out linkedEntrezId))
{
AddEnsemblOrphan(ensemblGene);
continue;
}
var refSeqGene = GeneUtilities.GetRefSeqGeneById(flatRefSeqGenes, linkedEntrezId);
if (refSeqGene == null)
{
AddEnsemblOrphan(ensemblGene);
continue;
}
// merge the Ensembl and RefSeq gene
var mergedGene = MutableGene.Clone(ensemblGene);
mergedGene.TranscriptDataSource = TranscriptDataSource.BothRefSeqAndEnsembl;
UpdateCoordinates(refSeqGene, mergedGene);
if (mergedGene.HgncId == -1 && refSeqGene.HgncId != -1)
{
mergedGene.HgncId = refSeqGene.HgncId;
}
mergedGene.EntrezGeneId = refSeqGene.EntrezGeneId;
_mergedGenes.Add(mergedGene);
refSeqGene.Invalid = true;
ensemblGene.Invalid = true;
_numMergedGenes++;
}
// add the unused RefSeq genes
foreach (var refSeqGene in flatRefSeqGenes)
{
if (refSeqGene.Invalid)
{
continue;
}
AddRefSeqOrphan(refSeqGene);
}
}
private void AddItem(string line, IDictionary <string, List <ISuppGeneItem> > geneAnnotations, bool skipGeneIdValidation, StreamWriter logWriter)
{
var splits = line.OptimizedSplit('\t');
if (splits.Length != _tags.Length)
{
throw new UserErrorException($"Column number mismatch!! Header has {_tags.Length} columns but {line} contains {splits.Length}");
}
string geneId = splits[1];
var annotationValues = new string[_numAnnotationColumns];
var hasAnnotation = false;
for (var i = 0; i < _numAnnotationColumns; i++)
{
string annotationValue = splits[i + NumRequiredColumns];
if (annotationValue != "" && annotationValue != ".")
{
hasAnnotation = true;
}
annotationValues[i] = annotationValue;
_annotationValidators[i](annotationValues[i], line);
}
if (!hasAnnotation)
{
throw new UserErrorException($"No annotation provided in line {line}");
}
string geneSymbol = GeneUtilities.GetGeneSymbolFromId(geneId, _entrezGeneIdToSymbol, _ensemblIdToSymbol);
if (geneSymbol == null)
{
if (!skipGeneIdValidation)
{
_unknownGenes.Add(geneId);
}
logWriter?.WriteLine($"Skipping unrecognized gene ID {geneId}");
return;
}
if (geneAnnotations.ContainsKey(geneSymbol))
{
throw new UserErrorException($"Found the same gene {geneSymbol} in different lines. Current line is: {line}");
}
geneAnnotations[geneSymbol] = new List <ISuppGeneItem> {
new CustomGene(geneSymbol, annotationValues.Select(x => new[] { x }).ToList(), JsonSchema, line)
};
}
private static ExitCodes ProgramExecution()
{
var(entrezGeneIdToSymbol, ensemblGeneIdToSymbol) = GeneUtilities.ParseUniversalGeneArchive(null, _universalGeneArchivePath);
string outputPrefix = GetOutputPrefix(_inputFile);
string ngaFilePath = Path.Combine(_outputDirectory, outputPrefix + SaCommon.NgaFileSuffix);
string ngaSchemaFilePath = ngaFilePath + SaCommon.JsonSchemaSuffix;
using (var parser = GeneAnnotationsParser.Create(GZipUtilities.GetAppropriateStreamReader(_inputFile), entrezGeneIdToSymbol, ensemblGeneIdToSymbol))
using (var ngaStream = FileUtilities.GetCreateStream(ngaFilePath))
using (var ngaWriter = CaUtilities.GetNgaWriter(ngaStream, parser, CaUtilities.GetInputFileName(_inputFile)))
using (var saJsonSchemaStream = FileUtilities.GetCreateStream(ngaSchemaFilePath))
using (var schemaWriter = new StreamWriter(saJsonSchemaStream))
{
ngaWriter.Write(parser.GetItems());
schemaWriter.Write(parser.JsonSchema);
}
return(ExitCodes.Success);
}
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);
}
private static ExitCodes ProgramExecution()
{
var(entrezGeneIdToSymbol, ensemblGeneIdToSymbol) = GeneUtilities.ParseUniversalGeneArchive(null, _universalGeneArchivePath);
string outputPrefix = GetOutputPrefix(_inputFile);
string ngaFilePath = Path.Combine(_outputDirectory, outputPrefix + SaCommon.GeneFileSuffix);
string ngaSchemaFilePath = ngaFilePath + SaCommon.JsonSchemaSuffix;
using (var parser = GeneAnnotationsParser.Create(GZipUtilities.GetAppropriateStreamReader(_inputFile), entrezGeneIdToSymbol, ensemblGeneIdToSymbol))
using (var ngaStream = FileUtilities.GetCreateStream(ngaFilePath))
using (var ngaWriter = CaUtilities.GetNgaWriter(ngaStream, parser, CaUtilities.GetInputFileName(_inputFile)))
using (var saJsonSchemaStream = FileUtilities.GetCreateStream(ngaSchemaFilePath))
using (var schemaWriter = new StreamWriter(saJsonSchemaStream))
{
ngaWriter.Write(parser.GetItems());
if (parser.GetUnknownGenes().Count > 0)
{
throw new UserErrorException($"The following gene IDs were not recognized in Nirvana: {string.Join(',',parser.GetUnknownGenes())}.");
}
schemaWriter.Write(parser.JsonSchema);
}
return(ExitCodes.Success);
}
private static GeneAnnotationsParser GetGeneAnnotationsParserFromCustomTsvStream(PersistentStream customTsvStream)
{
var(entrezGeneIdToSymbol, ensemblGeneIdToSymbol) = GeneUtilities.ParseUniversalGeneArchive(null, LambdaUrlHelper.GetUgaUrl());
return(GeneAnnotationsParser.Create(new StreamReader(GZipUtilities.GetAppropriateStream(customTsvStream)), entrezGeneIdToSymbol, ensemblGeneIdToSymbol));
}
