public IEnumerable <SpliceAiItem> GetItems()
{
var previousItems = new List <SpliceAiItem>();
using (var reader = FileUtilities.GetStreamReader(_stream))
{
string line;
while ((line = reader.ReadLine()) != null)
{
// Skip empty lines.
if (string.IsNullOrWhiteSpace(line))
{
continue;
}
// Skip comments.
if (line.OptimizedStartsWith('#'))
{
continue;
}
var item = ExtractItem(line);
if (item == null)
{
continue;
}
if (previousItems.Count == 0 || SpliceAiItem.CompareTo(item, previousItems[0]) == 0)
{
//starting or extending new position
previousItems.Add(item);
continue;
}
//performing sanity check
SanityCheck(previousItems);
UpdateGeneSymbols(previousItems);
foreach (var spliceAiItem in previousItems)
{
yield return(spliceAiItem);
}
previousItems.Clear();
previousItems.Add(item);
}
}
//clearing off the final items (they should all be at the same position
UpdateGeneSymbols(previousItems);
foreach (var spliceAiItem in previousItems)
{
yield return(spliceAiItem);
}
Console.WriteLine($"{_unresolvedSymbols.Count} unresolved gene symbols encountered. Symbols:");
foreach (var symbol in _unresolvedSymbols)
{
Console.Write(symbol + ',');
}
}
public static int CompareTo(SpliceAiItem one, SpliceAiItem other)
{
if (one.Chromosome.Index != other.Chromosome.Index)
{
return(one.Chromosome.Index.CompareTo(other.Chromosome.Index));
}
return(one.Position.CompareTo(other.Position));
}
private void UpdateGeneSymbol(SpliceAiItem item)
{
if (_spliceToNirvanaSymbols.TryGetValue(item.Hgnc, out var nirHgnc))
{
item.Hgnc = nirHgnc;
}
else
{
_unresolvedSymbols.Add(item.Hgnc);
}
}
private void UpdateGeneSymbol(SpliceAiItem item)
{
if (_geneTree == null || _geneSynonyms == null)
{
return;
}
var chromosome = item.Chromosome;
var position = item.Position;
if (_currentPositionGeneSymbols.Count > 1)
{
return; //for multiple genes for a position, we cannot update the symbol
}
var nirvanaGenes = _geneTree.GetAllOverlappingValues(chromosome.Index, position, position);
if (nirvanaGenes == null)
{
item.Hgnc = null;
return;
}
var uniqueOverlapping = new HashSet <string>(nirvanaGenes);
if (uniqueOverlapping.Contains(item.Hgnc))
{
return;
}
//gene not found in cache
if (uniqueOverlapping.Count == 1)
{
item.Hgnc = uniqueOverlapping.First(); //update gene symbol
}
else
{
if (!_geneSynonyms.TryGetValue(item.Hgnc, out var symbolsList))
{
return;
}
var commonSymbols = symbolsList.Intersect(uniqueOverlapping).ToArray();
if (commonSymbols.Length == 1)
{
item.Hgnc = commonSymbols[0];
}
else
{
_unresolvedSymbols.Add(item.Hgnc);
}
}
}
