public void BuildScaffold()
{
const int kmerLength = 6;
const int dangleThreshold = 3;
const int redundantThreshold = 7;
List <ISequence> 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();
Assert.AreEqual(contigs.Count, 10);
CloneLibrary.Instance.AddLibrary("abc", (float)5, (float)20);
GraphScaffoldBuilder scaffold = new GraphScaffoldBuilder();
IEnumerable <ISequence> scaffoldSeq = scaffold.BuildScaffold(
sequences, contigs, this.KmerLength, 3, 0);
Assert.AreEqual(scaffoldSeq.Count(), 8);
Assert.IsTrue(new string(scaffoldSeq.First().Select(a => (char)a).ToArray()).Equals(
"ATGCCTCCTATCTTAGCGCGC"));
scaffold.Dispose();
}
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 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", (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 = new string(contig.Select(a => (char)a).ToArray());
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 = new string(scaffold.Select(a => (char)a).ToArray());
Assert.IsTrue(
expectedScaffolds.Contains(scaffoldSeq) ||
expectedScaffolds.Contains(scaffoldSeq.GetReverseComplement(new char[scaffoldSeq.Length])));
}
}
}