/// <summary>
/// Map paired reads to contigs using FASTA sequence header.
/// </summary>
/// <param name="readContigMap">Map between reads and contigs</param>
/// <param name="reads">Sequences of reads.</param>
/// <returns>Contig Mate Pair map.</returns>
protected ContigMatePairs MapPairedReadsToContigs(ReadContigMap readContigMap, IList <ISequence> reads)
{
ContigMatePairs contigMatePairs = new ContigMatePairs();
contigMatePairs = _mapPairedReads.MapContigToMatePairs(reads, readContigMap);
return(contigMatePairs);
}
/// <summary>
/// Public method mapping Reads to Contigs.
/// </summary>
/// <param name="contigs">List of sequences of contigs.</param>
/// <param name="reads">List of input reads.</param>
/// <param name="kmerLength">Length of kmer.</param>
/// <returns>Contig Read Map.</returns>
public ReadContigMap Map(IList <ISequence> contigs, IList <ISequence> reads, int kmerLength)
{
KmerIndexerDictionary map = SequenceToKmerBuilder.BuildKmerDictionary(contigs, kmerLength);
ReadContigMap maps = new ReadContigMap();
using (ThreadLocal <char[]> rcBuilder = new ThreadLocal <char[]>(() => new char[kmerLength]))
{
Parallel.ForEach(reads, (ISequence readSequence) =>
{
IEnumerable <string> kmers = SequenceToKmerBuilder.GetKmerStrings(readSequence, kmerLength);
ReadIndex read = new ReadIndex(readSequence);
IList <KmerIndexer> positions;
foreach (string kmer in kmers)
{
if (map.TryGetValue(kmer, out positions) ||
map.TryGetValue(kmer.GetReverseComplement(rcBuilder.Value), out positions))
{
read.ContigReadMatchIndexes.Add(positions);
}
}
IList <Task <IList <ReadMap> > > tasks =
new List <Task <IList <ReadMap> > >();
//Stores information about contigs for which tasks has been generated.
IList <int> visitedContigs = new List <int>();
//Creates Task for every read in nodes for a given contig.
for (int index = 0; index < read.ContigReadMatchIndexes.Count; index++)
{
int readPosition = index;
foreach (KmerIndexer kmer in read.ContigReadMatchIndexes[index])
{
int contigIndex = kmer.SequenceIndex;
if (!visitedContigs.Contains(contigIndex))
{
visitedContigs.Add(contigIndex);
tasks.Add(Task <IList <ReadMap> > .Factory.StartNew(t =>
MapRead(readPosition, read.ContigReadMatchIndexes, contigIndex, read.ReadSequence.Count, kmerLength),
TaskCreationOptions.AttachedToParent));
}
}
}
Dictionary <ISequence, IList <ReadMap> > overlapMaps =
new Dictionary <ISequence, IList <ReadMap> >();
for (int index = 0; index < visitedContigs.Count; index++)
{
overlapMaps.Add(contigs[visitedContigs[index]], tasks[index].Result);
}
lock (maps)
{
if (!maps.ContainsKey(read.ReadSequence.DisplayID))
{
maps.Add(read.ReadSequence.DisplayID, overlapMaps);
}
else
{
throw new ArgumentException(
string.Format(CultureInfo.CurrentCulture, Resource.DuplicatingReadIds, read.ReadSequence.DisplayID));
}
}
});
}
return(maps);
}
/// <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="kmerLength">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(
IList <ISequence> reads,
IList <ISequence> contigs,
int kmerLength,
int depth = 10,
int redundancy = 2)
{
if (contigs == null)
{
throw new ArgumentNullException("contigs");
}
if (null == reads)
{
throw new ArgumentNullException("reads");
}
if (kmerLength <= 0)
{
throw new ArgumentException(Properties.Resource.KmerLength);
}
if (depth <= 0)
{
throw new ArgumentException(Resource.Depth);
}
if (redundancy < 0)
{
throw new ArgumentException(Resource.NegativeRedundancy);
}
_depth = depth;
_redundancy = redundancy;
_kmerLength = kmerLength;
IList <ISequence> readSeqs = reads.AsParallel().Where(s => s.All <ISequenceItem>(c => !c.IsAmbiguous && !c.IsGap)).ToList();
//Step1: Generate contig overlap graph.
DeBruijnGraph contigGraph = GenerateContigOverlapGraph(contigs);
IEnumerable <DeBruijnNode> nodes = contigGraph.Nodes.Where(t => t.ExtensionsCount == 0);
foreach (DeBruijnNode node in nodes)
{
contigs.Remove(contigGraph.GetNodeSequence(node));
}
// Step2: Map Reads to contigs.
ReadContigMap readContigMap = ReadContigMap(contigs, readSeqs);
contigs = null;
// Step3: Generate Contig Mate Pair Map.
ContigMatePairs contigMatePairs = MapPairedReadsToContigs(readContigMap, readSeqs);
readContigMap = 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));
}