public void DistanceCalculationwithTwoContigs()
{
const int KmerLength = 6;
IList <ISequence> sequences = new List <ISequence>()
{
new Sequence(Alphabets.DNA, "GATCTGATAA")
{
ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor substrate 1 (IRS1) on chromosome 2.X1:0.5K"
},
new Sequence(Alphabets.DNA, "ATCTGATAAG")
{
ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor on chromosome 2.F:0.5K"
},
new Sequence(Alphabets.DNA, "TCTGATAAGG")
{
ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor on chromosome 2.2:0.5K"
},
new Sequence(Alphabets.DNA, "TTTTTGATGG")
{
ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor substrate 1 (IRS1) on chromosome 2.Y1:0.5K"
},
new Sequence(Alphabets.DNA, "TTTTGATGGC")
{
ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor on chromosome 2.R:0.5K"
},
new Sequence(Alphabets.DNA, "TTTGATGGCA")
{
ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor on chromosome 2.1:0.5K"
}
};
IList <ISequence> contigs = new List <ISequence> {
new Sequence(Alphabets.DNA, "GATCTGATAAGG"),
new Sequence(Alphabets.DNA, "TTTTTGATGGCA")
};
ReadContigMapper mapper = new ReadContigMapper();
ReadContigMap maps = mapper.Map(contigs, sequences, KmerLength);
MatePairMapper mapPairedReads = new MatePairMapper();
ContigMatePairs pairs = mapPairedReads.MapContigToMatePairs(sequences, maps);
OrientationBasedMatePairFilter filter = new OrientationBasedMatePairFilter();
ContigMatePairs contigpairedReads = filter.FilterPairedReads(pairs);
DistanceCalculator calc = new DistanceCalculator(contigpairedReads);
contigpairedReads = calc.CalculateDistance();
Assert.AreEqual(contigpairedReads.Values.Count, 1);
Assert.IsTrue(contigpairedReads.ContainsKey(contigs[0]));
Dictionary <ISequence, IList <ValidMatePair> > map = contigpairedReads[contigs[0]];
Assert.IsTrue(map.ContainsKey(contigs[1]));
IList <ValidMatePair> valid = map[contigs[1]];
Assert.AreEqual(valid.First().DistanceBetweenContigs[0], (float)478.000031);
Assert.AreEqual(valid.First().DistanceBetweenContigs[1], (float)477.0);
Assert.AreEqual(valid.First().StandardDeviation[0], (float)14.1421356);
Assert.AreEqual(valid.First().StandardDeviation[1], (float)14.1421356);
Assert.AreEqual(valid.First().Weight, 2);
}
public void DistanceCalculationwithTwoContigsWeightedMean()
{
const int KmerLength = 6;
List <ISequence> sequences = new List <ISequence>();
Sequence seq = new Sequence(Alphabets.DNA, "GATCTGATAA".Select(a => (byte)a).ToArray());
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "substrate 1 (IRS1) on chromosome 2.x1:2K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ATCTGATAAG".Select(a => (byte)a).ToArray());
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "on chromosome 2.f:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TCTGATAAGG".Select(a => (byte)a).ToArray());
seq.ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor"
+ "on chromosome 2.2:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTTGATGG".Select(a => (byte)a).ToArray());
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "substrate 1 (IRS1) on chromosome 2.y1:2K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTGATGGC".Select(a => (byte)a).ToArray());
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "on chromosome 2.r:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTGATGGCA".Select(a => (byte)a).ToArray());
seq.ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor"
+ "on chromosome 2.1:0.5K";
sequences.Add(seq);
IList <ISequence> contigs = new List <ISequence> {
new Sequence(Alphabets.DNA, "GATCTGATAAGG".Select(a => (byte)a).ToArray()),
new Sequence(Alphabets.DNA, "TTTTTGATGGCA".Select(a => (byte)a).ToArray())
};
ReadContigMapper mapper = new ReadContigMapper();
ReadContigMap maps = mapper.Map(contigs, sequences, KmerLength);
MatePairMapper mapPairedReads = new MatePairMapper();
ContigMatePairs pairedReads = mapPairedReads.MapContigToMatePairs(sequences, maps);
OrientationBasedMatePairFilter filter = new OrientationBasedMatePairFilter();
ContigMatePairs contigpairedReads = filter.FilterPairedReads(pairedReads);
DistanceCalculator calc = new DistanceCalculator(contigpairedReads);
contigpairedReads = calc.CalculateDistance();
Assert.AreEqual(contigpairedReads.Values.Count, 1);
Assert.IsTrue(contigpairedReads.ContainsKey(contigs[0]));
Dictionary <ISequence, IList <ValidMatePair> > map = contigpairedReads[contigs[0]];
Assert.IsTrue(map.ContainsKey(contigs[1]));
IList <ValidMatePair> valid = map[contigs[1]];
Assert.AreEqual(valid.First().DistanceBetweenContigs[0], (float)1228.0);
Assert.AreEqual(valid.First().DistanceBetweenContigs[1], (float)1227.0);
Assert.AreEqual(valid.First().StandardDeviation[0], (float)60);
Assert.AreEqual(valid.First().StandardDeviation[1], (float)60);
Assert.AreEqual(valid.First().Weight, 2);
}
public void FilterMatePairWithTwoContigs()
{
const int kmerLength = 6;
IList <ISequence> sequences = new List <ISequence>();
Sequence seq = new Sequence(Alphabets.DNA, "GATCTGATAA");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor substrate 1 (IRS1) on chromosome 2.X1:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ATCTGATAAG");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor on chromosome 2.F:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TCTGATAAGG");
seq.ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor on chromosome 2.2:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTTGATGG");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor substrate 1 (IRS1) on chromosome 2.Y1:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTGATGGC");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor on chromosome 2.R:0.5K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTGATGGCA");
seq.ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor on chromosome 2.1:0.5K";
sequences.Add(seq);
IList <ISequence> contigs = new List <ISequence> {
new Sequence(Alphabets.DNA, "GATCTGATAAGG"),
new Sequence(Alphabets.DNA, "TTTTTGATGGCA")
};
ReadContigMapper mapper = new ReadContigMapper();
ReadContigMap maps = mapper.Map(contigs, sequences, kmerLength);
MatePairMapper mapPairedReads = new MatePairMapper();
ContigMatePairs pairs = mapPairedReads.MapContigToMatePairs(sequences, maps);
OrientationBasedMatePairFilter filter = new OrientationBasedMatePairFilter();
ContigMatePairs contigpairedReads = filter.FilterPairedReads(pairs);
Assert.AreEqual(contigpairedReads.Values.Count, 1);
Assert.IsTrue(contigpairedReads.ContainsKey(contigs[0]));
Dictionary <ISequence, IList <ValidMatePair> > map = contigpairedReads[contigs[0]];
Assert.IsTrue(map.ContainsKey(contigs[1]));
List <ValidMatePair> valid = Sort(map[contigs[1]], sequences);
Assert.AreEqual(valid[0].ForwardReadStartPosition[0], 1);
Assert.AreEqual(valid[0].ReverseReadReverseComplementStartPosition[0], 10);
Assert.AreEqual(valid[0].ReverseReadStartPosition[0], 10);
Assert.AreEqual(valid[1].ForwardReadStartPosition[0], 0);
Assert.AreEqual(valid[1].ReverseReadReverseComplementStartPosition[0], 11);
Assert.AreEqual(valid[1].ReverseReadStartPosition[0], 9);
}
public void ContigPairReadMap()
{
const int kmerLength = 6;
IList <ISequence> readSeqs = new List <ISequence>();
Sequence read = new Sequence(Alphabets.DNA, "GATCTGATAA");
read.DisplayID = "0.x1:abc";
readSeqs.Add(read);
read = new Sequence(Alphabets.DNA, "ATCTGATAAG");
read.DisplayID = "1.F:abc";
readSeqs.Add(read);
read = new Sequence(Alphabets.DNA, "TCTGATAAGG");
read.DisplayID = "2.2:abc";
readSeqs.Add(read);
read = new Sequence(Alphabets.DNA, "TTTTTGATGG");
read.DisplayID = "0.y1:abc";
readSeqs.Add(read);
read = new Sequence(Alphabets.DNA, "TTTTGATGGC");
read.DisplayID = "1.R:abc";
readSeqs.Add(read);
read = new Sequence(Alphabets.DNA, "TTTGATGGCA");
read.DisplayID = "2.1:abc";
readSeqs.Add(read);
IList <ISequence> contigs = new List <ISequence> {
new Sequence(Alphabets.DNA, "GATCTGATAAGG"),
new Sequence(Alphabets.DNA, "TTTTTGATGGCA")
};
ReadContigMapper mapper = new ReadContigMapper();
ReadContigMap maps = mapper.Map(contigs, readSeqs, kmerLength);
MatePairMapper pair = new MatePairMapper();
ContigMatePairs map = pair.MapContigToMatePairs(readSeqs, maps);
Assert.AreEqual(map.Count, 2);
Dictionary <ISequence, IList <ValidMatePair> > reverseContigs;
Assert.IsTrue(map.TryGetValue(contigs[0], out reverseContigs));
Assert.AreEqual(reverseContigs.Count, 1);
IList <ValidMatePair> matePairs;
Assert.IsTrue(reverseContigs.TryGetValue(contigs[1], out matePairs));
Assert.AreEqual(matePairs.Count, 2);
Assert.AreEqual(matePairs[0].ForwardReadStartPosition.First(), 0);
Assert.AreEqual(matePairs[0].ReverseReadStartPosition.First(), 9);
Assert.AreEqual(matePairs[1].ForwardReadStartPosition.First(), 1);
Assert.AreEqual(matePairs[1].ReverseReadStartPosition.First(), 10);
Assert.IsTrue(map.TryGetValue(contigs[1], out reverseContigs));
Assert.AreEqual(reverseContigs.Count, 1);
Assert.IsTrue(reverseContigs.TryGetValue(contigs[0], out matePairs));
Assert.AreEqual(matePairs.Count, 1);
Assert.AreEqual(matePairs[0].ForwardReadStartPosition.First(), 2);
Assert.AreEqual(matePairs[0].ReverseReadStartPosition.First(), 11);
}
public void MatePairMapper1()
{
IList <ISequence> sequences = new List <ISequence>();
IList <MatePair> pairedreads = new List <MatePair>();
Sequence seq = new Sequence(Alphabets.DNA, "GGGAAAAATCAGATT");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "substrate 1 (IRS1) on chromosome 2.X1:2K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ACGTGTGTGTCCCCC");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "on chromosome 2.X1:10K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ACGTGTGKAAAAAAA");
seq.ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor"
+ "on chromosome 2.X1:50K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "GGGAAAAATCAGATT");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "substrate 1 (IRS1) on chromosome 2.Y1:2K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ACGTGTGTGTCCCCC");
seq.ID = ">gi|263191773|ref|NG_015830.1| H**o sapiens insulin receptor"
+ "on chromosome 2.Y1:10K";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ACGTGTGKAAAAAAA");
seq.ID = ">gi|263191773|ref | H**o sapiens ........insulin receptor"
+ "on chromosome 2.Y1:50K";
sequences.Add(seq);
MatePairMapper pair = new MatePairMapper();
pairedreads = Sort(pair.Map(sequences), sequences);
Assert.AreEqual(pairedreads.Count, 3);
ISequence expectedForward = new Sequence(Alphabets.DNA, "ACGTGTGKAAAAAAA");
ISequence expectedReverse = new Sequence(Alphabets.DNA, "ACGTGTGKAAAAAAA");
string lib = "50K";
Assert.AreEqual(pairedreads[0].GetForwardRead(sequences).ToString(), expectedForward.ToString());
Assert.AreEqual(pairedreads[0].GetReverseRead(sequences).ToString(), expectedReverse.ToString());
Assert.AreEqual(pairedreads[0].Library, lib);
Assert.AreEqual(pairedreads[0].MeanLengthOfLibrary, 65000);
Assert.AreEqual(pairedreads[0].StandardDeviationOfLibrary, 13334);
expectedForward = new Sequence(Alphabets.DNA, "ACGTGTGTGTCCCCC");
expectedReverse = new Sequence(Alphabets.DNA, "ACGTGTGTGTCCCCC");
lib = "10K";
Assert.AreEqual(pairedreads[1].GetForwardRead(sequences).ToString(), expectedForward.ToString());
Assert.AreEqual(pairedreads[1].GetReverseRead(sequences).ToString(), expectedReverse.ToString());
Assert.AreEqual(pairedreads[1].Library, lib);
Assert.AreEqual(pairedreads[1].MeanLengthOfLibrary, 10000);
Assert.AreEqual(pairedreads[1].StandardDeviationOfLibrary, 1000);
expectedForward = new Sequence(Alphabets.DNA, "GGGAAAAATCAGATT");
expectedReverse = new Sequence(Alphabets.DNA, "GGGAAAAATCAGATT");
lib = "2K";
Assert.AreEqual(pairedreads[2].GetForwardRead(sequences).ToString(), expectedForward.ToString());
Assert.AreEqual(pairedreads[2].GetReverseRead(sequences).ToString(), expectedReverse.ToString());
Assert.AreEqual(pairedreads[2].Library, lib);
Assert.AreEqual(pairedreads[2].MeanLengthOfLibrary, 2000);
Assert.AreEqual(pairedreads[2].StandardDeviationOfLibrary, 100);
}
public void TracePathTestWithPalindromicContig()
{
const int kmerLength = 6;
const int dangleThreshold = 3;
const int redundantThreshold = 7;
List <ISequence> sequences = new List <ISequence>();
Sequence seq = new Sequence(Alphabets.DNA, "ATGCCTC");
seq.DisplayID = ">10.x1:abc";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "CCTCCTAT");
seq.DisplayID = "1";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TCCTATC");
seq.DisplayID = "2";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TGCCTCCT");
seq.DisplayID = "3";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "ATCTTAGC");
seq.DisplayID = "4";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "CTATCTTAG");
seq.DisplayID = "5";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "CTTAGCG");
seq.DisplayID = "6";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "GCCTCCTAT");
seq.DisplayID = ">8.x1:abc";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TAGCGCGCTA");
seq.DisplayID = ">8.y1:abc";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "AGCGCGC");
seq.DisplayID = ">9.x1:abc";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTTT");
seq.DisplayID = "7";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTTAAA");
seq.DisplayID = "8";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TAAAAA");
seq.DisplayID = "9";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTTAG");
seq.DisplayID = "10";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "TTTAGC");
seq.DisplayID = "11";
sequences.Add(seq);
seq = new Sequence(Alphabets.DNA, "GCGCGCCGCGCG");
seq.DisplayID = "12";
sequences.Add(seq);
KmerLength = kmerLength;
SequenceReads.Clear();
AddSequenceReads(sequences);
CreateGraph();
DanglingLinksThreshold = dangleThreshold;
DanglingLinksPurger = new DanglingLinksPurger(dangleThreshold);
RedundantPathLengthThreshold = redundantThreshold;
RedundantPathsPurger = new RedundantPathsPurger(redundantThreshold);
UnDangleGraph();
RemoveRedundancy();
IList <ISequence> contigs = BuildContigs();
ReadContigMapper mapper = new ReadContigMapper();
ReadContigMap maps = mapper.Map(contigs, sequences, kmerLength);
MatePairMapper builder = new MatePairMapper();
CloneLibrary.Instance.AddLibrary("abc", (float)5, (float)15);
ContigMatePairs pairedReads = builder.MapContigToMatePairs(sequences, maps);
ContigMatePairs overlap;
OrientationBasedMatePairFilter filter = new OrientationBasedMatePairFilter();
overlap = filter.FilterPairedReads(pairedReads, 0);
DistanceCalculator dist = new DistanceCalculator();
dist.CalculateDistance(overlap);
Graph.BuildContigGraph(contigs, this.KmerLength);
TracePath path = new TracePath();
IList <ScaffoldPath> paths = path.FindPaths(Graph, overlap, kmerLength, 3);
Assert.AreEqual(paths.Count, 3);
Assert.AreEqual(paths.First().Count, 3);
ScaffoldPath scaffold = paths.First();
DeBruijnGraph graph = Graph;
Assert.IsTrue(graph.GetNodeSequence(scaffold[0].Key).ToString().Equals("ATGCCTCCTATCTTAGC"));
Assert.IsTrue(graph.GetNodeSequence(scaffold[1].Key).ToString().Equals("TTAGCGCG"));
Assert.IsTrue(graph.GetNodeSequence(scaffold[2].Key).ToString().Equals("GCGCGC"));
}
public void TracePathTestWithPalindromicContig()
{
const int kmerLengthConst = 5;
const int dangleThreshold = 3;
const int redundantThreshold = 6;
var sequences = new List <ISequence>()
{
new Sequence(Alphabets.DNA, "ATGCCTC")
{
ID = "0"
},
new Sequence(Alphabets.DNA, "CCTCCTAT")
{
ID = "1"
},
new Sequence(Alphabets.DNA, "TCCTATC")
{
ID = "2"
},
new Sequence(Alphabets.DNA, "TGCCTCCT")
{
ID = "3"
},
new Sequence(Alphabets.DNA, "ATCTTAGC")
{
ID = "4"
},
new Sequence(Alphabets.DNA, "CTATCTTAG")
{
ID = "5"
},
new Sequence(Alphabets.DNA, "CTTAGCG")
{
ID = "6"
},
new Sequence(Alphabets.DNA, "GCCTCCTAT")
{
ID = "7"
},
new Sequence(Alphabets.DNA, "TAGCGCGCTA")
{
ID = "8"
},
new Sequence(Alphabets.DNA, "AGCGCGC")
{
ID = "9"
},
new Sequence(Alphabets.DNA, "TTTTTT")
{
ID = "10"
},
new Sequence(Alphabets.DNA, "TTTTTAAA")
{
ID = "11"
},
new Sequence(Alphabets.DNA, "TAAAAA")
{
ID = "12"
},
new Sequence(Alphabets.DNA, "TTTTAG")
{
ID = "13"
},
new Sequence(Alphabets.DNA, "TTTAGC")
{
ID = "14"
},
new Sequence(Alphabets.DNA, "GCGCGCCGCGCG")
{
ID = "15"
},
};
KmerLength = kmerLengthConst;
SequenceReads.Clear();
SetSequenceReads(sequences);
CreateGraph();
DanglingLinksThreshold = dangleThreshold;
DanglingLinksPurger = new DanglingLinksPurger(dangleThreshold);
RedundantPathLengthThreshold = redundantThreshold;
RedundantPathsPurger = new RedundantPathsPurger(redundantThreshold);
UnDangleGraph();
RemoveRedundancy();
IList <ISequence> contigs = BuildContigs().ToList();
ReadContigMapper mapper = new ReadContigMapper();
ReadContigMap maps = mapper.Map(contigs, sequences, kmerLengthConst);
MatePairMapper builder = new MatePairMapper();
CloneLibrary.Instance.AddLibrary("abc", 5, 15);
ContigMatePairs pairedReads = builder.MapContigToMatePairs(sequences, maps);
OrientationBasedMatePairFilter filter = new OrientationBasedMatePairFilter();
ContigMatePairs overlap = filter.FilterPairedReads(pairedReads, 0);
DistanceCalculator dist = new DistanceCalculator(overlap);
overlap = dist.CalculateDistance();
ContigGraph graph = new ContigGraph();
graph.BuildContigGraph(contigs, this.KmerLength);
TracePath path = new TracePath();
IList <ScaffoldPath> paths = path.FindPaths(graph, overlap, kmerLengthConst, 3);
Assert.AreEqual(paths.Count, 3);
Assert.AreEqual(paths.First().Count, 3);
ScaffoldPath scaffold = paths.First();
Assert.AreEqual("ATGCCTCCTATCTTAGC", graph.GetNodeSequence(scaffold[0].Key).ConvertToString());
Assert.AreEqual("TTAGCGCG", graph.GetNodeSequence(scaffold[1].Key).ConvertToString());
Assert.AreEqual("GCGCGC", graph.GetNodeSequence(scaffold[2].Key).ConvertToString());
}