private IEnumerable <GnomadItem> GetRemainingItems(IEnumerator <GnomadItem> enumerator)
{
do
{
var item = enumerator.Current;
if (item == null)
{
yield break;
}
if (item.AllAlleleNumber == null || item.AllAlleleNumber.Value == 0)
{
continue;
}
yield return(GnomadUtilities.GetNormalizedItem(item, _sequenceProvider));
} while (enumerator.MoveNext());
}
/// <summary>
/// Merging genomic an exomic items to create one stream of gnomad entries
/// </summary>
/// <returns></returns>
public IEnumerable <GnomadItem> GetCombinedItems()
{
using (var genomeEnumerator = GetItems(_genomeReader, GnomadDataType.Genome).GetEnumerator())
using (var exomeEnumerator = GetItems(_exomeReader, GnomadDataType.Exome).GetEnumerator())
{
var hasGenomicItem = genomeEnumerator.MoveNext();
var hasExomeItem = exomeEnumerator.MoveNext();
var minHeap = new MinHeap <GnomadItem>(GnomadItem.CompareTo);
while (hasExomeItem && hasGenomicItem)
{
var genomeItem = genomeEnumerator.Current;
var exomeItem = exomeEnumerator.Current;
var position = Math.Min(genomeItem.Position, exomeItem.Position);
while (hasGenomicItem && genomeItem.Position == position)
{
//all items for a position should be gathered so as to resolve conflicts properly
minHeap.Add(GnomadUtilities.GetNormalizedItem(genomeItem, _sequenceProvider));
hasGenomicItem = genomeEnumerator.MoveNext();
genomeItem = genomeEnumerator.Current;
}
while (hasExomeItem && exomeItem.Position == position)
{
minHeap.Add(GnomadUtilities.GetNormalizedItem(exomeItem, _sequenceProvider));
hasExomeItem = exomeEnumerator.MoveNext();
exomeItem = exomeEnumerator.Current;
}
// at this point, the min heap should not be empty
int heapPosition = minHeap.GetMin().Position;
while (minHeap.Count() > 0 && heapPosition < position - VariantUtils.MaxUpstreamLength)
{
var(genomeItems, exomeItems) = GetMinItems(minHeap);
foreach (var item in GnomadUtilities.GetMergedItems(genomeItems, exomeItems).Values)
{
if (item.AllAlleleNumber == null || item.AllAlleleNumber.Value == 0)
{
continue;
}
yield return(item);
}
}
}
//flush out the last positions in heap
while (minHeap.Count() > 0)
{
var(genomeItems, exomeItems) = GetMinItems(minHeap);
foreach (var item in GnomadUtilities.GetMergedItems(genomeItems, exomeItems).Values)
{
yield return(item);
}
}
//now, only one of the iterator is left
if (hasGenomicItem)
{
foreach (var item in GetRemainingItems(genomeEnumerator))
{
yield return(item);
}
}
if (hasExomeItem)
{
foreach (var item in GetRemainingItems(exomeEnumerator))
{
yield return(item);
}
}
}
}
