public void BuildScaffold()
{
const int kmerLength = 6;
const int dangleThreshold = 3;
const int redundantThreshold = 7;
List <ISequence> sequences = TestInputs.GetReadsForScaffolds();
KmerLength = kmerLength;
SequenceReads.Clear();
this.AddSequenceReads(sequences);
CreateGraph();
DanglingLinksThreshold = dangleThreshold;
DanglingLinksPurger = new DanglingLinksPurger(dangleThreshold);
RedundantPathLengthThreshold = redundantThreshold;
RedundantPathsPurger = new RedundantPathsPurger(redundantThreshold);
UnDangleGraph();
RemoveRedundancy();
IList <ISequence> contigs = BuildContigs();
DeBruijnGraph graph = Graph;
CloneLibrary.Instance.AddLibrary("abc", (float)5, (float)20);
GraphScaffoldBuilder scaffold = new GraphScaffoldBuilder();
IList <ISequence> scaffoldSeq = scaffold.BuildScaffold(
sequences, contigs, this.KmerLength, 3, 0);
Assert.AreEqual(scaffoldSeq.Count, 8);
Assert.IsTrue(scaffoldSeq[0].ToString().Equals(
"ATGCCTCCTATCTTAGCGCGC"));
}
public void AssemblerTestWithScaffoldBuilder()
{
const int kmerLength = 6;
const int dangleThreshold = 3;
const int redundantThreshold = 7;
ParallelDeNovoAssembler assembler = new ParallelDeNovoAssembler();
assembler.KmerLength = kmerLength;
assembler.DanglingLinksThreshold = dangleThreshold;
assembler.RedundantPathLengthThreshold = redundantThreshold;
assembler.ScaffoldRedundancy = 0;
assembler.Depth = 3;
CloneLibrary.Instance.AddLibrary("abc", (float)5, (float)20);
PaDeNAAssembly result = (PaDeNAAssembly)assembler.Assemble(TestInputs.GetReadsForScaffolds(), true);
Assert.AreEqual(10, result.ContigSequences.Count);
HashSet <string> expectedContigs = new HashSet <string>
{
"GCGCGC",
"TTTTTT",
"TTTTTA",
"TTTTAA",
"TTTAAA",
"ATGCCTCCTATCTTAGC",
"TTTTAGC",
"TTAGCGCG",
"CGCGCCGCGC",
"CGCGCG"
};
foreach (ISequence contig in result.ContigSequences)
{
string contigSeq = contig.ToString();
Assert.IsTrue(
expectedContigs.Contains(contigSeq) ||
expectedContigs.Contains(contigSeq.GetReverseComplement(new char[contigSeq.Length])));
}
Assert.AreEqual(8, result.Scaffolds.Count);
HashSet <string> expectedScaffolds = new HashSet <string>
{
"ATGCCTCCTATCTTAGCGCGC",
"TTTTTT",
"TTTTTA",
"TTTTAA",
"TTTAAA",
"CGCGCCGCGC",
"TTTTAGC",
"CGCGCG"
};
foreach (ISequence scaffold in result.Scaffolds)
{
string scaffoldSeq = scaffold.ToString();
Assert.IsTrue(
expectedScaffolds.Contains(scaffoldSeq) ||
expectedScaffolds.Contains(scaffoldSeq.GetReverseComplement(new char[scaffoldSeq.Length])));
}
}