public void BuildScaffold()
{
const int kmerLength = 6;
const int dangleThreshold = 3;
const int redundantThreshold = 7;
var sequences = new List <ISequence>(TestInputs.GetReadsForScaffolds());
KmerLength = kmerLength;
SequenceReads.Clear();
this.SetSequenceReads(sequences);
CreateGraph();
DanglingLinksThreshold = dangleThreshold;
DanglingLinksPurger = new DanglingLinksPurger(dangleThreshold);
RedundantPathLengthThreshold = redundantThreshold;
RedundantPathsPurger = new RedundantPathsPurger(redundantThreshold);
UnDangleGraph();
RemoveRedundancy();
IList <ISequence> contigs = BuildContigs().ToList();
CloneLibrary.Instance.AddLibrary("abc", 5, 20);
using (GraphScaffoldBuilder scaffold = new GraphScaffoldBuilder())
{
IEnumerable <ISequence> scaffoldSeq = scaffold.BuildScaffold(sequences, contigs, this.KmerLength, 3, 0);
HashSet <string> expected = new HashSet <string>
{
"ATGCCTCCTATCTTAGCGCGC", "CGCGCCGCGC", "TTTTTT", "CGCGCG", "TTTTAGC", "TTTTTA", "TTTAAA", "TTTTAA",
};
AlignmentHelpers.CompareSequenceLists(expected, scaffoldSeq);
}
}
public void AssemblerTestWithScaffoldBuilder()
{
const int kmerLength = 6;
const int dangleThreshold = 3;
const int redundantThreshold = 7;
using (ParallelDeNovoAssembler assembler = new ParallelDeNovoAssembler())
{
assembler.KmerLength = kmerLength;
assembler.DanglingLinksThreshold = dangleThreshold;
assembler.RedundantPathLengthThreshold = redundantThreshold;
assembler.ScaffoldRedundancy = 0;
assembler.Depth = 3;
CloneLibrary.Instance.AddLibrary("abc", 5, 20);
PadenaAssembly result = (PadenaAssembly)assembler.Assemble(TestInputs.GetReadsForScaffolds(), true);
HashSet <string> expectedContigs = new HashSet <string>
{
"TTTTTT", "CGCGCG", "TTAGCGCG", "CGCGCCGCGC", "GCGCGC", "TTTTTA", "TTTTAA", "TTTAAA", "TTTTAGC", "ATGCCTCCTATCTTAGC"
};
AlignmentHelpers.CompareSequenceLists(expectedContigs, result.ContigSequences);
HashSet <string> expectedScaffolds = new HashSet <string>
{
"ATGCCTCCTATCTTAGCGCGC", "TTTAAA", "TTTTTT", "TTTTAGC", "TTTTAA", "CGCGCCGCGC", "TTTTTA", "CGCGCG"
};
AlignmentHelpers.CompareSequenceLists(expectedScaffolds, result.Scaffolds);
}
}
public void TestDanglingLinksPurger()
{
const int KmerLength = 11;
const int DangleThreshold = 3;
List <ISequence> readSeqs = TestInputs.GetDanglingReads();
SequenceReads.Clear();
this.SetSequenceReads(readSeqs);
this.KmerLength = KmerLength;
DanglingLinksThreshold = DangleThreshold;
DanglingLinksPurger = new DanglingLinksPurger(DangleThreshold);
CreateGraph();
long graphCount = Graph.NodeCount;
long graphEdges = Graph.GetNodes().Select(n => n.ExtensionsCount).Sum();
var graphNodes = Graph.GetNodes().Select(n => Graph.GetNodeSequence(n)).ToList();
DanglingLinksThreshold = DangleThreshold;
UnDangleGraph();
long dangleRemovedGraphCount = Graph.NodeCount;
long dangleRemovedGraphEdge = Graph.GetNodes().Select(n => n.ExtensionsCount).Sum();
var dangleRemovedGraphNodes = Graph.GetNodes().Select(n => Graph.GetNodeSequence(n)).ToList();
// Compare the two graphs
Assert.AreEqual(2, graphCount - dangleRemovedGraphCount);
Assert.AreEqual(4, graphEdges - dangleRemovedGraphEdge);
var checkList = graphNodes.Except(dangleRemovedGraphNodes, new SequenceEqualityComparer());
HashSet <string> expected = new HashSet <string> {
"ATCGAACGATG", "TCGAACGATGA"
};
AlignmentHelpers.CompareSequenceLists(expected, checkList);
}