/** Tests the adding of a sequence list to the histogram */
public void TestAddSequenceList()
{
SequenceList list = new SequenceList();
list.Add(new Sequence(AlphabetType.DNA, "actga"));
list.Add(new Sequence(AlphabetType.DNA, "ctaca"));
histo.Add(list);
Assert.AreEqual(4, histo.HistoValue(alpha['a']));
Assert.AreEqual(3, histo.HistoValue(alpha['c']));
Assert.AreEqual(2, histo.HistoValue(alpha['t']));
Assert.AreEqual(1, histo.HistoValue(alpha['g']));
}
/** Tests constructor */
public void TestConstructor()
{
SequenceList list = new SequenceList();
list.Add(new Sequence(AlphabetType.DNA, "aa", false));
list.Add( new Sequence( AlphabetType.DNA, "at", false ) );
Block block = new Block("test", list, null, 0.0);
Assert.AreEqual(1.000, block.Get('a', 0), 1e-3);
Assert.AreEqual(-0.584, block.Get('c', 0), 1e-3);
Assert.AreEqual(-0.584, block.Get('t', 0), 1e-3);
Assert.AreEqual(-0.584, block.Get('g', 0), 1e-3);
Assert.AreEqual(0.415, block.Get('a', 1), 1e-3);
Assert.AreEqual(-0.584, block.Get('c', 1), 1e-3);
Assert.AreEqual(0.415, block.Get('t', 1), 1e-3);
Assert.AreEqual(-0.584, block.Get('g', 1), 1e-3);
}
/// <summary>
/// Reads the Genbank file and have it parsed by MBF library.
/// </summary>
/// <param name="genbankFileURL">Your genbank file path</param>
/// <returns></returns>
private SequenceList ParseSequencePath
(string genbankFileURL)
{
if (IsOnline)
throw new NotImplementedException
("online genbank reading is not supported in this version!");
//Download the file and parse it
//Create the parser first
ISequenceParser gbParser = new GenBankParser();
//Always Try parsing multi sequence in a file
List<ISequence> mbfSequences = gbParser.Parse(genbankFileURL);
SequenceList bioSeqList = new SequenceList();
foreach (Sequence mbfseq in mbfSequences)
{
ConvertToBioPatMLSeq(mbfseq);
bioSeqList.Add(ConvertToBioPatMLSeq(mbfseq));
}
return bioSeqList;
}
/// <summary>
/// Reads in the fasta file.
/// </summary>
/// <param name="reader">your local filepath for genbank</param>
/// <returns>list of BioPatML Sequences</returns>
public override SequenceList Read(TextReader reader)
{
//Create the parser first
ISequenceParser fastaParser = new FastaParser();
List<ISequence> mbfSequences = fastaParser.Parse(reader);
SequenceList bioSeqList = new SequenceList();
foreach (Sequence mbfseq in mbfSequences)
{
bioSeqList.Add(ConvertToBioPatMLSeq(mbfseq));
}
return bioSeqList;
}
/// <summary>
/// The param could also be a stringreader.
/// </summary>
/// <param name="reader"></param>
/// <returns></returns>
private SequenceList ParseSequencePath
(TextReader reader)
{
//Create the parser first
ISequenceParser gbParser = new GenBankParser();
//Always Try parsing multi sequence in a reader
List<ISequence> mbfSequences = gbParser.Parse(reader);
SequenceList bioSeqList = new SequenceList();
foreach (Sequence mbfseq in mbfSequences)
{
ConvertToBioPatMLSeq(mbfseq);
bioSeqList.Add(ConvertToBioPatMLSeq(mbfseq));
}
return bioSeqList;
}
/// <summary>
/// Reads the parameters and populate the attributes for this pattern.
/// </summary>
/// <exception cref="System.ArgumentNullException">
/// Thrown when sequences in blocks are missing.</exception>
/// <param name="node"></param>
/// <param name="definition">The Definition element where the node sits in</param>
public override void ReadNode
(XmlNode node, Definition definition)
{
PatternName = (XMLHelper.GetAttrValueString(node, "name"));
Threshold = (XMLHelper.GetAttrValDouble(node, "threshold"));
Impact = (XMLHelper.GetAttrValDouble(node, "impact"));
PWMalphabet = AlphabetFactory.Instance
(XMLHelper.GetAttrValueString(node, "alphabet"));
SequenceList seqList = new SequenceList();
node = node.FirstChild;
while (node != null)
{
if (node.Name.Equals("Sequence"))
{
String letters = node.InnerText.Trim();
if (letters == null)
throw new ArgumentNullException
("Sequences in Block are missing!");
seqList.Add(new Sequence(PWMalphabet, letters, false));
}
node = node.NextSibling;
}
Estimate(seqList, null);
}
