private static void CombineSet(ICollection <UgaGene> combinedGenes, IEnumerable <UgaGene> uga37,
IEnumerable <UgaGene> uga38, ICollection <UgaGene> remainingGenes37, ICollection <UgaGene> remainingGenes38)
{
var keyToGene37 = uga37.GetMultiValueDict(GetKey);
var keyToGene38 = uga38.GetMultiValueDict(GetKey);
var keys = GetAllKeys(keyToGene37.Keys, keyToGene38.Keys);
foreach (var key in keys)
{
var genes37 = GetGenesByKey(keyToGene37, key);
var genes38 = GetGenesByKey(keyToGene38, key);
CombinerUtils.RemoveGenes(genes37, remainingGenes37);
CombinerUtils.RemoveGenes(genes38, remainingGenes38);
// this happens for both Entrez Gene Only & Ensembl Only
if (genes37.Count == 1 && genes38.Count == 1)
{
var gene37 = genes37[0];
var gene38 = genes38[0];
var mergedGene = CombinerUtils.Merge(gene37, gene38);
combinedGenes.Add(mergedGene);
continue;
}
// the following situations happen if we have:
// - one gene from GRCh37 and none from GRCh38 (or vice versa)
// - two or more non-overlapping genes on the same assembly (14 occurrences)
CombinerUtils.AddOrphans(combinedGenes, genes37);
CombinerUtils.AddOrphans(combinedGenes, genes38);
}
}
public void Combine(List <UgaGene> combinedGenes, HashSet <UgaGene> remainingGenes37,
HashSet <UgaGene> remainingGenes38)
{
var hgncIds = GetHgncIds(remainingGenes37, remainingGenes38);
var genesByHgnc37 = remainingGenes37.GetMultiValueDict(x => x.HgncId);
var genesByHgnc38 = remainingGenes38.GetMultiValueDict(x => x.HgncId);
foreach (var hgncId in hgncIds)
{
var genes37 = GetGenesByHgncId(genesByHgnc37, hgncId);
var genes38 = GetGenesByHgncId(genesByHgnc38, hgncId);
CombinerUtils.RemoveGenes(genes37, remainingGenes37);
CombinerUtils.RemoveGenes(genes38, remainingGenes38);
// merge if we have one gene on each genome assembly and they're on the same strand
if (genes37.Count == 1 && genes38.Count == 1)
{
var gene37 = genes37[0];
var gene38 = genes38[0];
if (gene37.OnReverseStrand == gene38.OnReverseStrand)
{
var mergedGene = CombinerUtils.Merge(gene37, gene38);
combinedGenes.Add(mergedGene);
continue;
}
}
// the following situations happen if we have:
// - one gene from GRCh37 and none from GRCh38 (or vice versa)
// - there is a mixture of genes forward and reverse strands (13 occurrences)
CombinerUtils.AddOrphans(combinedGenes, genes37);
CombinerUtils.AddOrphans(combinedGenes, genes38);
}
}