/// <summary>
/// Distance calculator.
/// </summary>
/// <param name="contigPairedReads">Contig pair reads.</param>
public DistanceCalculator(ContigMatePairs contigPairedReads)
{
if (contigPairedReads == null)
{
throw new ArgumentNullException("contigPairedReads");
}
this.contigPairedReads = contigPairedReads;
}
/// <summary>
/// Performs Breadth First Search to traverse through graph to generate scaffold paths.
/// </summary>
/// <param name="overlapGraph">Contig Overlap Graph.</param>
/// <param name="contigPairedReadMaps">InterContig Distances.</param>
/// <param name="lengthOfKmer">Length of Kmer.</param>
/// <param name="searchDepth">Depth to which graph is searched.</param>
/// <returns>List of paths/scaffold.</returns>
public IList <ScaffoldPath> FindPaths(
ContigGraph overlapGraph,
ContigMatePairs contigPairedReadMaps,
int lengthOfKmer,
int searchDepth = 10)
{
if (overlapGraph == null)
{
throw new ArgumentNullException("deBruijnGraph");
}
if (contigPairedReadMaps == null)
{
throw new ArgumentNullException("contigPairedReadMaps");
}
if (lengthOfKmer <= 0)
{
throw new ArgumentException(Resource.KmerLength);
}
if (searchDepth <= 0)
{
throw new ArgumentException(Resource.Depth);
}
this.graph = overlapGraph;
this.kmerLength = lengthOfKmer;
this.depth = searchDepth;
List <ScaffoldPath> scaffoldPaths = new List <ScaffoldPath>();
Parallel.ForEach(
overlapGraph.Nodes,
(Node node) =>
{
Dictionary <ISequence, IList <ValidMatePair> > contigPairedReadMap;
if (contigPairedReadMaps.TryGetValue(overlapGraph.GetNodeSequence(node), out contigPairedReadMap))
{
List <ScaffoldPath> scaffoldPath = TraverseGraph(node, contigPairedReadMap);
lock (scaffoldPaths)
{
scaffoldPaths.AddRange(scaffoldPath);
}
}
});
return(scaffoldPaths);
}
/// <summary>
/// Calculate distance between contigs using paired reads.
/// </summary>
/// <param name="contigMatePairs">Contig Mate Pair map.</param>
/// <returns>Number of contig-read pairs.</returns>
protected int CalculateDistanceBetweenContigs(ContigMatePairs contigMatePairs)
{
if (contigMatePairs == null)
{
throw new ArgumentNullException("contigMatePairs");
}
if (this.distanceCalculator == null)
{
this.distanceCalculator = new DistanceCalculator(contigMatePairs);
contigMatePairs = this.distanceCalculator.CalculateDistance();
}
else
{
contigMatePairs = this.distanceCalculator.CalculateDistance();
}
// this dictionary is updated in this step.
return(contigMatePairs.Count);
}
/// <summary>
/// Filter Paired Read Based on Orientation.
/// </summary>
/// <param name="matePairMap">Map between contigs using mate pair information.</param>
/// <param name="redundancy">Number of mate pairs required to create a link between two contigs.
/// Hierarchical Scaffolding With Bambus
/// by: Mihai Pop, Daniel S. Kosack, Steven L. Salzberg
/// Genome Research, Vol. 14, No. 1. (January 2004), pp. 149-159.</param>
public ContigMatePairs FilterPairedReads(ContigMatePairs matePairMap, int redundancy = 2)
{
if (null == matePairMap)
{
throw new ArgumentNullException("matePairMap");
}
if (redundancy < 0)
{
throw new ArgumentException(Bio.Algorithms.Assembly.Padena.Properties.Resource.NegativeRedundancy);
}
foreach (KeyValuePair <ISequence, Dictionary <ISequence, IList <ValidMatePair> > > matePair in matePairMap)
{
foreach (KeyValuePair <ISequence, IList <ValidMatePair> > validMatePair in matePair.Value)
{
if (matePair.Key != validMatePair.Key)
{
Dictionary <ISequence, IList <ValidMatePair> > validMatePairs;
if (matePairMap.TryGetValue(validMatePair.Key, out validMatePairs))
{
IList <ValidMatePair> pair;
if (validMatePairs.TryGetValue(matePair.Key, out pair))
{
OrientationFilter(pair, validMatePair.Value, redundancy);
}
else
{
if (validMatePair.Value.Count < redundancy)
{
validMatePair.Value.Clear();
}
}
}
else
{
if (validMatePair.Value.Count < redundancy)
{
validMatePair.Value.Clear();
}
}
}
else
{
validMatePair.Value.Clear();
}
}
}
ContigMatePairs newMatePairMap = new ContigMatePairs();
Parallel.ForEach(
matePairMap,
(KeyValuePair <ISequence, Dictionary <ISequence, IList <ValidMatePair> > > matePair) =>
{
Dictionary <ISequence, IList <ValidMatePair> > map = new Dictionary <ISequence, IList <ValidMatePair> >();
foreach (KeyValuePair <ISequence, IList <ValidMatePair> > validMatePair in matePair.Value)
{
if (validMatePair.Value.Count > 0)
{
map.Add(validMatePair.Key, validMatePair.Value);
}
}
if (map.Count > 0)
{
lock (newMatePairMap)
{
newMatePairMap.Add(matePair.Key, map);
}
}
});
return(newMatePairMap);
}
/// <summary>
/// Performs Breadth First Search in contig overlap graph.
/// </summary>
/// <param name="contigGraph">Contig Graph.</param>
/// <param name="contigMatePairs">Contig Mate Pair map.</param>
/// <returns>List of Scaffold Paths.</returns>
protected IList <ScaffoldPath> TracePath(ContigGraph contigGraph, ContigMatePairs contigMatePairs)
{
return(this.tracePath.FindPaths(contigGraph, contigMatePairs, this.kmerLength, this.depthField));
}
/// <summary>
/// Filter reads based on orientation of contigs.
/// </summary>
/// <param name="contigMatePairs">Contig Mate Pair map.</param>
/// <returns>Returns Contig Mate Pair map.</returns>
protected ContigMatePairs FilterReadsBasedOnOrientation(ContigMatePairs contigMatePairs)
{
return(this.pairedReadFilter.FilterPairedReads(contigMatePairs, this.redundancyField));
}
/// <summary>
/// Builds scaffolds from list of reads and contigs.
/// </summary>
/// <param name="reads">List of reads.</param>
/// <param name="contigs">List of contigs.</param>
/// <param name="lengthofKmer">Kmer Length.</param>
/// <param name="depth">Depth for graph traversal.</param>
/// <param name="redundancy">Number of mate pairs required to create a link between two contigs.
/// Hierarchical Scaffolding With Bambus
/// by: Mihai Pop, Daniel S. Kosack, Steven L. Salzberg
/// Genome Research, Vol. 14, No. 1. (January 2004), pp. 149-159.</param>
/// <returns>List of scaffold sequences.</returns>
public IList <ISequence> BuildScaffold(
IEnumerable <ISequence> reads,
IList <ISequence> contigs,
int lengthofKmer,
int depth = 10,
int redundancy = 2)
{
if (contigs == null)
{
throw new ArgumentNullException("contigs");
}
if (null == reads)
{
throw new ArgumentNullException("reads");
}
if (lengthofKmer <= 0)
{
throw new ArgumentException(Properties.Resource.KmerLength);
}
if (depth <= 0)
{
throw new ArgumentException(Resource.Depth);
}
if (redundancy < 0)
{
throw new ArgumentException(Resource.NegativeRedundancy);
}
this.depthField = depth;
this.redundancyField = redundancy;
this.kmerLength = lengthofKmer;
IEnumerable <ISequence> readSeqs = ValidateReads(reads);
//Step1: Generate contig overlap graph.
IList <ISequence> contigsList = new List <ISequence>(contigs);
ContigGraph contigGraph = GenerateContigOverlapGraph(contigsList);
IEnumerable <Node> nodes = contigGraph.Nodes.Where(t => t.ExtensionsCount == 0);
foreach (Node node in nodes)
{
contigsList.Remove(contigGraph.GetNodeSequence(node));
}
// Step2: Map Reads to contigs.
ReadContigMap readContigMaps = ReadContigMap(contigsList, readSeqs);
contigsList = null;
// Step3: Generate Contig Mate Pair Map.
ContigMatePairs contigMatePairs = MapPairedReadsToContigs(readContigMaps, readSeqs);
readContigMaps = null;
// Step4: Filter Paired Reads.
contigMatePairs = FilterReadsBasedOnOrientation(contigMatePairs);
// Step5: Distance Calculation.
CalculateDistanceBetweenContigs(contigMatePairs);
// Step6: Trace Scaffold Paths.
IList <ScaffoldPath> paths = TracePath(contigGraph, contigMatePairs);
contigMatePairs = null;
// Step7: Assemble paths.
PathPurger(paths);
// Step8: Generate sequence of scaffolds.
return(GenerateScaffold(contigGraph, paths));
}