public void ValidateDnaSparseSequenceConstAlpIndexByteList()
{
byte[] byteArrayObj = encodingObj.GetBytes("AGCT");
IEnumerable <byte> seqItems =
new List <Byte>()
{
byteArrayObj[0], byteArrayObj[1], byteArrayObj[2], byteArrayObj[3]
};
SparseSequence sparseSeq = new SparseSequence(Alphabets.DNA, 4, seqItems);
Assert.IsNotNull(sparseSeq);
Assert.IsNotNull(sparseSeq.Statistics);
Assert.AreEqual(8, sparseSeq.Count);
SequenceStatistics seqStatObj = sparseSeq.Statistics;
Assert.AreEqual(1, seqStatObj.GetCount('A'));
Assert.AreEqual(1, seqStatObj.GetCount('G'));
Assert.AreEqual(1, seqStatObj.GetCount('C'));
Assert.AreEqual(1, seqStatObj.GetCount('T'));
Console.WriteLine("SparseSequence BVT: Validation of SparseSequence(alp, index, seq items) constructor is completed");
ApplicationLog.WriteLine("SparseSequence BVT: Validation of SparseSequence(alp, index, seq items) constructor is completed");
}
/// <summary>
/// Calculate GC content for a read
/// </summary>
/// <param name="seqStats">SequenceStatistics object</param>
/// <param name="length">The length of the read</param>
/// <returns>GC percentage of a read</returns>
public double GCContentBySequence(SequenceStatistics seqStats, long length)
{
if (seqStats == null)
{
throw new ArgumentNullException("seqStats");
}
return(100 * (double)(seqStats.GetCount('G') + seqStats.GetCount('C') + seqStats.GetCount('S')) / length);
}
public void CreateStatsWithMixedcaseLetterSequence()
{
ISequence sequence = new Sequence(Alphabets.DNA, "aAaAaAaA");
SequenceStatistics stats = new SequenceStatistics(sequence);
Assert.AreEqual(8, stats.GetCount('A'));
Assert.AreEqual(8, stats.GetCount(65));
Assert.AreEqual(8, stats.GetCount('a'));
Assert.AreEqual(1.0, stats.GetFraction('A'));
Assert.AreEqual(1.0, stats.GetFraction(65));
Assert.AreEqual(1.0, stats.GetFraction('a'));
}
public void CreateSimpleStatsWithSingleLetterSequence()
{
ISequence sequence = new Sequence(Alphabets.DNA, "A");
SequenceStatistics stats = new SequenceStatistics(sequence);
Assert.AreEqual(1, stats.GetCount('A'));
Assert.AreEqual(1, stats.GetCount(65));
Assert.AreEqual(0, stats.GetCount('C'));
Assert.AreEqual(0, stats.GetCount('G'));
Assert.AreEqual(0, stats.GetCount('T'));
Assert.AreEqual(1.0, stats.GetFraction('A'));
Assert.AreEqual(1.0, stats.GetFraction(65));
}
public void CreateSimpleStatsWithSingleLetterSequence()
{
ISequence sequence = new Sequence(Alphabets.DNA, "A");
SequenceStatistics stats = new SequenceStatistics(sequence);
Assert.AreEqual(1, stats.GetCount('A'));
Assert.AreEqual(1, stats.GetCount(65));
Assert.AreEqual(0, stats.GetCount('C'));
Assert.AreEqual(0, stats.GetCount('G'));
Assert.AreEqual(0, stats.GetCount('T'));
Assert.AreEqual(1.0, stats.GetFraction('A'));
Assert.AreEqual(1.0, stats.GetFraction(65));
}
static void DumpStatsForOneSequence()
{
ISequence oneSequence = null;
var parser = SequenceParsers.FindParserByFileName(Filename);
if (parser != null)
{
oneSequence = parser.Parse().First();
parser.Close();
}
if (oneSequence == null)
{
Console.WriteLine("Could not load sequence.");
return;
}
SequenceStatistics stats =
new SequenceStatistics(oneSequence);
foreach (var symbol in oneSequence.Alphabet)
{
Console.WriteLine("{0} = Count={1}, Fraction={2}",
(char)symbol, stats.GetCount(symbol),
stats.GetFraction(symbol));
}
}
public void ValidateDnaSparseSequenceConstAlpIndexByte()
{
byte[] byteArrayObj = Encoding.ASCII.GetBytes("AGCT");
var sparseSeq = new SparseSequence(Alphabets.DNA, 1, byteArrayObj[0]);
Assert.IsNotNull(sparseSeq);
Assert.IsNotNull(sparseSeq.Statistics);
SequenceStatistics seqStatObj = sparseSeq.Statistics;
Assert.AreEqual(1, seqStatObj.GetCount('A'));
ApplicationLog.WriteLine("SparseSequence BVT: Validation of SparseSequence(alp, index, byte) constructor is completed");
}
/// <summary>
/// When implemented in a derived class, performs the execution of the activity.
/// </summary>
/// <param name="context">The execution context under which the activity executes.</param>
protected override string Execute(CodeActivityContext context)
{
var sequence = Sequence.Get(context);
StringBuilder buff = new StringBuilder();
buff.AppendLine(sequence.ID);
buff.Append("Statistics: ");
buff.Append(sequence.Count);
buff.Append(" Total");
var statistics = new SequenceStatistics(sequence);
foreach (var symbol in sequence.Alphabet)
{
buff.AppendFormat(" - {0}:", (char)symbol);
buff.Append(statistics.GetCount(symbol));
}
buff.AppendLine();
buff.AppendLine();
for (int i = 0; i < sequence.Count; i++)
{
if ((i % 50) == 0)
{
string num = (i + 1).ToString(CultureInfo.InvariantCulture);
int pad = 5 - num.Length;
StringBuilder buff2 = new StringBuilder();
for (int j = 0; j < pad; j++)
{
buff2.Append(' ');
}
buff2.Append(num);
buff.Append(buff2);
}
if ((i % 10) == 0)
{
buff.Append(' ');
}
buff.Append((char)sequence[i]);
if ((i % 50) == 49)
{
buff.AppendLine();
}
}
buff.AppendLine();
return(buff.ToString());
}
static void Main(string[] args)
{
var parser = SequenceParsers.FindParserByFileName(Filename);
if (parser == null)
{
Console.WriteLine("No parser found.");
return;
}
List <ISequence> allSequences = parser
.Parse()
.ToList();
IAlphabet alphabet = allSequences[0].Alphabet;
long totalColums = allSequences.Min(s => s.Count);
//byte[] data = new byte[allSequences.Count];
for (int column = 0; column < totalColums; column++)
{
// for (int row = 0; row < allSequences.Count; row++)
// data[row] = allSequences[row][column];
//ISequence newSequence = new Sequence(alphabet, data);
ISequence newSequence = new Sequence(AmbiguousRnaAlphabet.Instance,
allSequences
.Select(s => s[column]).ToArray());
SequenceStatistics stats =
new SequenceStatistics(newSequence);
var TopCount =
alphabet
.Where(symbol => !alphabet.CheckIsGap(symbol))
.Select(symbol =>
new
{
Symbol = (char)symbol,
Count = stats.GetCount(symbol),
Frequency = stats.GetFraction(symbol)
})
.Where(tc => tc.Count > 0)
.OrderByDescending(c => c.Count)
.FirstOrDefault();
if (TopCount != null)
{
Console.WriteLine("{0}: {1} = {2} of {3}",
column, TopCount.Symbol, TopCount.Count, totalColums);
}
}
}
public void ValidateProteinSparseSequenceConstAlpIndexByte()
{
byte[] byteArrayObj = encodingObj.GetBytes("KIEG");
SparseSequence sparseSeq = new SparseSequence(Alphabets.Protein, 1, byteArrayObj[0]);
Assert.IsNotNull(sparseSeq);
Assert.IsNotNull(sparseSeq.Statistics);
SequenceStatistics seqStatObj = sparseSeq.Statistics;
Assert.AreEqual(1, seqStatObj.GetCount('K'));
Console.WriteLine("SparseSequence P1: Validation of SparseSequence(alp, index, byte) constructor is completed");
ApplicationLog.WriteLine("SparseSequence P1: Validation of SparseSequence(alp, index, byte) constructor is completed");
}
public void ValidateProteinSparseSequenceConstAlpIndexByteList()
{
byte[] byteArrayObj = Encoding.ASCII.GetBytes("KIEG");
IEnumerable <byte> seqItems =
new List <Byte> {
byteArrayObj[0], byteArrayObj[1], byteArrayObj[2], byteArrayObj[3]
};
var sparseSeq = new SparseSequence(Alphabets.Protein, 4, seqItems);
Assert.IsNotNull(sparseSeq);
Assert.IsNotNull(sparseSeq.Statistics);
Assert.AreEqual(8, sparseSeq.Count);
SequenceStatistics seqStatObj = sparseSeq.Statistics;
Assert.AreEqual(1, seqStatObj.GetCount('K'));
Assert.AreEqual(1, seqStatObj.GetCount('I'));
Assert.AreEqual(1, seqStatObj.GetCount('E'));
Assert.AreEqual(1, seqStatObj.GetCount('G'));
ApplicationLog.WriteLine("SparseSequence P1: Validation of SparseSequence(alp, index, seq items) constructor is completed");
}
public void CreateStatsAndConsumeEnumerable()
{
ISequence sequence = new Sequence(Alphabets.DNA, "ACGT--ACGT--ACGT--");
SequenceStatistics stats = new SequenceStatistics(sequence);
int loopCounts = 0;
foreach (var value in stats.SymbolCounts)
{
Assert.AreEqual(value.Item2, stats.GetCount(value.Item1));
loopCounts++;
}
Assert.AreEqual(5, loopCounts);
}
public void CreateStatsWithSeveralMixedcaseLetterSequence()
{
ISequence sequence = new Sequence(Alphabets.DNA, "a-c-g-t-A-C-G-T-");
SequenceStatistics stats = new SequenceStatistics(sequence);
Assert.AreEqual(2, stats.GetCount('A'));
Assert.AreEqual(2, stats.GetCount('a'));
Assert.AreEqual(2, stats.GetCount('C'));
Assert.AreEqual(2, stats.GetCount('c'));
Assert.AreEqual(2, stats.GetCount('G'));
Assert.AreEqual(2, stats.GetCount('g'));
Assert.AreEqual(2, stats.GetCount('T'));
Assert.AreEqual(2, stats.GetCount('t'));
Assert.AreEqual(8, stats.GetCount('-'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('A'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('C'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('G'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('T'));
Assert.AreEqual(.5, stats.GetFraction('-'));
Assert.AreEqual(.5, stats.GetFraction(45));
}
/// <summary>
/// The execution method for the activity.
/// </summary>
/// <param name="executionContext">The execution context.</param>
/// <returns>The execution status.</returns>
protected override ActivityExecutionStatus Execute(ActivityExecutionContext executionContext)
{
StringBuilder buff = new StringBuilder();
buff.AppendLine(Sequence.ID);
buff.Append("Statistics: ");
buff.Append(Sequence.Count);
buff.Append(" Total");
var statistics = new SequenceStatistics(Sequence);
if (Sequence.Alphabet == Alphabets.DNA)
{
buff.Append(" - G: ");
buff.Append(statistics.GetCount(Alphabets.DNA.G));
buff.Append(" - A: ");
buff.Append(statistics.GetCount(Alphabets.DNA.A));
buff.Append(" - T: ");
buff.Append(statistics.GetCount(Alphabets.DNA.T));
buff.Append(" - C: ");
buff.Append(statistics.GetCount(Alphabets.DNA.C));
}
else if (Sequence.Alphabet == Alphabets.RNA)
{
buff.Append(" - G: ");
buff.Append(statistics.GetCount(Alphabets.RNA.G));
buff.Append(" - A: ");
buff.Append(statistics.GetCount(Alphabets.RNA.A));
buff.Append(" - U: ");
buff.Append(statistics.GetCount(Alphabets.RNA.U));
buff.Append(" - C: ");
buff.Append(statistics.GetCount(Alphabets.RNA.C));
}
buff.AppendLine();
buff.AppendLine();
for (int i = 0; i < Sequence.Count; i++)
{
if ((i % 50) == 0)
{
string num = (i + 1).ToString();
int pad = 5 - num.Length;
StringBuilder buff2 = new StringBuilder();
for (int j = 0; j < pad; j++)
{
buff2.Append(' ');
}
buff2.Append(num);
buff.Append(buff2.ToString());
}
if ((i % 10) == 0)
{
buff.Append(' ');
}
buff.Append((char)Sequence[i]);
if ((i % 50) == 49)
{
buff.AppendLine();
}
}
buff.AppendLine();
Data = buff.ToString();
return(ActivityExecutionStatus.Closed);
}
public void CreateStatsWithSeveralMixedcaseLetterSequence()
{
ISequence sequence = new Sequence(Alphabets.DNA, "a-c-g-t-A-C-G-T-");
SequenceStatistics stats = new SequenceStatistics(sequence);
Assert.AreEqual(2, stats.GetCount('A'));
Assert.AreEqual(2, stats.GetCount('a'));
Assert.AreEqual(2, stats.GetCount('C'));
Assert.AreEqual(2, stats.GetCount('c'));
Assert.AreEqual(2, stats.GetCount('G'));
Assert.AreEqual(2, stats.GetCount('g'));
Assert.AreEqual(2, stats.GetCount('T'));
Assert.AreEqual(2, stats.GetCount('t'));
Assert.AreEqual(8, stats.GetCount('-'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('A'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('C'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('G'));
Assert.AreEqual(2.0 / 16.0, stats.GetFraction('T'));
Assert.AreEqual(.5, stats.GetFraction('-'));
Assert.AreEqual(.5, stats.GetFraction(45));
}
public void CreateStatsAndConsumeEnumerable()
{
ISequence sequence = new Sequence(Alphabets.DNA, "ACGT--ACGT--ACGT--");
SequenceStatistics stats = new SequenceStatistics(sequence);
int loopCounts = 0;
foreach (var value in stats.SymbolCounts)
{
Assert.AreEqual(value.Item2, stats.GetCount(value.Item1));
loopCounts++;
}
Assert.AreEqual(5,loopCounts);
}
public void CreateStatsWithMixedcaseLetterSequence()
{
ISequence sequence = new Sequence(Alphabets.DNA, "aAaAaAaA");
SequenceStatistics stats = new SequenceStatistics(sequence);
Assert.AreEqual(8, stats.GetCount('A'));
Assert.AreEqual(8, stats.GetCount(65));
Assert.AreEqual(8, stats.GetCount('a'));
Assert.AreEqual(1.0, stats.GetFraction('A'));
Assert.AreEqual(1.0, stats.GetFraction(65));
Assert.AreEqual(1.0, stats.GetFraction('a'));
}