// ReSharper disable once InconsistentNaming
public void SRGAP2C()
{
var genesA = GetSrgap2CEnsemblGenes();
var genesB = GetSrgap2CRefSeqGenes();
var linkedEnsemblIds = new Dictionary <string, string> {
["ENSG00000171943"] = "653464"
};
var merger = new GeneMerger(genesA, genesB, linkedEnsemblIds);
var mergedGenes = merger.Merge();
Assert.Equal(2, mergedGenes.Count);
var mergedGene = mergedGenes[0];
Assert.Equal(0, mergedGene.ReferenceIndex);
Assert.Equal(120835810, mergedGene.Start);
Assert.Equal(120838261, mergedGene.End);
Assert.Equal("SRGAP2C", mergedGene.Symbol);
Assert.Equal(-1, mergedGene.HgncId);
Assert.Equal("653464", mergedGene.EntrezGeneId.ToString());
Assert.True(mergedGene.EnsemblId.IsEmpty);
Assert.True(mergedGene.OnReverseStrand);
var mergedGene2 = mergedGenes[1];
Assert.Equal(0, mergedGene2.ReferenceIndex);
Assert.Equal(121107124, mergedGene2.Start);
Assert.Equal(121131061, mergedGene2.End);
Assert.Equal("SRGAP2C", mergedGene2.Symbol);
Assert.Equal(30584, mergedGene2.HgncId);
Assert.Equal("653464", mergedGene2.EntrezGeneId.ToString());
Assert.Equal("ENSG00000171943", mergedGene2.EnsemblId.ToString());
Assert.False(mergedGene2.OnReverseStrand);
}
// ReSharper disable once InconsistentNaming
public void CDRT1()
{
var genesA = GetCdrt1EnsemblGenes();
var genesB = GetCdrt1RefSeqGenes();
var linkedEnsemblIds = new Dictionary <string, string> {
["ENSG00000241322"] = "374286"
};
var merger = new GeneMerger(genesA, genesB, linkedEnsemblIds);
var mergedGenes = merger.Merge();
Assert.Equal(3, mergedGenes.Count);
var mergedGene = mergedGenes[0];
Assert.Equal(16, mergedGene.ReferenceIndex);
Assert.Equal(15468797, mergedGene.Start);
Assert.Equal(15469590, mergedGene.End);
Assert.Equal("CDRT1", mergedGene.Symbol);
Assert.Equal(-1, mergedGene.HgncId);
Assert.True(mergedGene.EntrezGeneId.IsEmpty);
Assert.Equal("ENSG00000181464", mergedGene.EnsemblId.ToString());
Assert.True(mergedGene.OnReverseStrand);
var mergedGene2 = mergedGenes[1];
Assert.Equal(16, mergedGene2.ReferenceIndex);
Assert.Equal(15468798, mergedGene2.Start);
Assert.Equal(15523018, mergedGene2.End);
Assert.Equal("CDRT1", mergedGene2.Symbol);
Assert.Equal(14379, mergedGene2.HgncId);
Assert.Equal("374286", mergedGene2.EntrezGeneId.ToString());
Assert.Equal("ENSG00000241322", mergedGene2.EnsemblId.ToString());
Assert.True(mergedGene2.OnReverseStrand);
var mergedGene3 = mergedGenes[2];
Assert.Equal(16, mergedGene3.ReferenceIndex);
Assert.Equal(15474805, mergedGene3.Start);
Assert.Equal(15554967, mergedGene3.End);
Assert.Equal("CDRT1", mergedGene3.Symbol);
Assert.Equal(-1, mergedGene3.HgncId);
Assert.True(mergedGene3.EntrezGeneId.IsEmpty);
Assert.Equal("ENSG00000251537", mergedGene3.EnsemblId.ToString());
Assert.True(mergedGene3.OnReverseStrand);
}
// ReSharper disable once InconsistentNaming
public void SH3RF3()
{
var genesA = GetSh3EnsemblGenes();
var genesB = GetSh3RefSeqGenes();
var linkedEnsemblIds = new Dictionary <string, string> {
["ENSG00000172985"] = "344558"
};
var merger = new GeneMerger(genesA, genesB, linkedEnsemblIds);
var mergedGenes = merger.Merge();
Assert.Equal(1, mergedGenes.Count);
var mergedGene = mergedGenes[0];
Assert.Equal(1, mergedGene.ReferenceIndex);
Assert.Equal(109745804, mergedGene.Start);
Assert.Equal(110262207, mergedGene.End);
Assert.Equal("SH3RF3", mergedGene.Symbol);
Assert.Equal(24699, mergedGene.HgncId);
Assert.Equal("344558", mergedGene.EntrezGeneId.ToString());
Assert.Equal("ENSG00000172985", mergedGene.EnsemblId.ToString());
Assert.False(mergedGene.OnReverseStrand);
}
/// <summary>
/// constructor
/// </summary>
public GeneCombiner(string inputGenesPath, string inputGenes2Path, List <string> geneInfoPaths, string hgncPath,
string refSeqGff3Path)
{
_geneInfoSource = ParseGeneInfoFiles(geneInfoPaths);
var entrezGeneIdToEnsemblId = new Dictionary <string, UniqueString>();
var ensemblIdToEntrezGeneId = new Dictionary <string, UniqueString>();
_hgncSource = ParseHgncFile(hgncPath, entrezGeneIdToEnsemblId, ensemblIdToEntrezGeneId);
Console.WriteLine();
Console.WriteLine("- linking Ensembl and Entrez gene IDs: ");
var linkedEnsemblIds = LinkIds(entrezGeneIdToEnsemblId, ensemblIdToEntrezGeneId);
Console.WriteLine();
Console.WriteLine("- loading RefSeq GFF3: ");
_refSeqGff3GeneInfo = GetRefSeqGff3GeneInfo(refSeqGff3Path);
Console.WriteLine();
Console.WriteLine("- loading genes: ");
string descriptionA = Path.GetFileName(inputGenesPath);
string descriptionB = Path.GetFileName(inputGenes2Path);
var genesA = LoadGenes(inputGenesPath, descriptionA);
var genesB = LoadGenes(inputGenes2Path, descriptionB);
Console.WriteLine();
Console.WriteLine("- update gene symbols: ");
var updaterA = new GeneSymbolUpdater(genesA, descriptionA, _geneInfoSource, _hgncSource);
updaterA.Update();
var updaterB = new GeneSymbolUpdater(genesB, descriptionB, _geneInfoSource, _hgncSource);
updaterB.Update();
Console.WriteLine();
Console.WriteLine("- flattening genes: ");
var flattenerA = new GeneFlattener(genesA, descriptionA);
var flatGenesA = flattenerA.Flatten();
var flattenerB = new GeneFlattener(genesB, descriptionB);
var flatGenesB = flattenerB.Flatten();
Console.WriteLine();
Console.WriteLine("- merging Ensembl and RefSeq:");
var merger = new GeneMerger(flatGenesA, flatGenesB, linkedEnsemblIds);
_mergedGenes = merger.Merge();
Console.WriteLine();
Console.WriteLine("- update HGNC ids:");
UpdateHgncIds(_mergedGenes);
}