private static IEnumerable <GeneInfo> LoadGeneInfoGenes(string filePath)
{
GeneInfo[] genes;
using (var streamReader = GZipUtilities.GetAppropriateStreamReader(filePath))
using (var reader = new GeneInfoReader(streamReader)) genes = reader.GetGenes();
return(genes);
}
private static SymbolDataSource ParseGeneInfoFiles(List <string> geneInfoPaths)
{
Console.WriteLine("- loading gene_info files:");
var entrezGeneIdToSymbol = new Dictionary <string, UniqueString>();
var ensemblIdToSymbol = new Dictionary <string, UniqueString>();
var entrezGeneIdToHgncId = new Dictionary <string, UniqueInt>();
var ensemblIdToHgncId = new Dictionary <string, UniqueInt>();
foreach (var geneInfoPath in geneInfoPaths)
{
Console.Write(" - {0}... ", Path.GetFileName(geneInfoPath));
int numEntries = 0;
using (var reader = new GeneInfoReader(geneInfoPath))
{
while (true)
{
var geneinfo = reader.Next();
if (geneinfo == null)
{
break;
}
if (geneinfo.IsEmpty)
{
continue;
}
numEntries++;
bool hasEntrezGeneId = !string.IsNullOrEmpty(geneinfo.EntrezGeneId);
bool hasEnsemblId = !string.IsNullOrEmpty(geneinfo.EnsemblId);
bool hasSymbol = !string.IsNullOrEmpty(geneinfo.Symbol);
bool hasHgncId = geneinfo.HgncId != -1;
if (hasSymbol)
{
if (hasEntrezGeneId)
{
AddIdToUniqueString(entrezGeneIdToSymbol, geneinfo.EntrezGeneId, geneinfo.Symbol);
}
if (hasEnsemblId)
{
AddIdToUniqueString(ensemblIdToSymbol, geneinfo.EnsemblId, geneinfo.Symbol);
}
}
if (hasHgncId)
{
if (hasEntrezGeneId)
{
AddIdToHgncId(entrezGeneIdToHgncId, geneinfo.EntrezGeneId, geneinfo.HgncId);
}
if (hasEnsemblId)
{
AddIdToHgncId(ensemblIdToHgncId, geneinfo.EnsemblId, geneinfo.HgncId);
}
}
}
}
Console.WriteLine($"{numEntries} entries loaded.");
}
Console.WriteLine($" - Entrez Gene ID -> symbol: {entrezGeneIdToSymbol.Count} ({GetNonConflictCount(entrezGeneIdToSymbol)})");
Console.WriteLine($" - Ensembl ID -> symbol: {ensemblIdToSymbol.Count} ({GetNonConflictCount(ensemblIdToSymbol)})");
Console.WriteLine($" - Entrez Gene ID -> HGNC id: {entrezGeneIdToHgncId.Count} ({GetNonConflictCount(entrezGeneIdToHgncId)})");
Console.WriteLine($" - Ensembl ID -> HGNC id: {ensemblIdToHgncId.Count} ({GetNonConflictCount(ensemblIdToHgncId)})");
return(new SymbolDataSource(entrezGeneIdToSymbol, ensemblIdToSymbol, entrezGeneIdToHgncId, ensemblIdToHgncId));
}