public void SetUp()
{
this.BiopatMLFilePath = string.Empty;
this.BioList = null;
this.MyPatterns = null;
SearchPosition = 1;
}
/// <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;
}
public void TestMotifPattern_Motif () {
BiopatMLFilePath = "BioPaperTestData/MotifPattern/Motif.xml";
using ( BioPatMBF_Reader gbReader = new BioPatMBF_Reader() ) {
BioList = gbReader.Read( Global.GetResourceReader( _singleDnaSeqGenBankFilename ) );
}
MyPatterns = DefinitionIO.Read( Global.GetResourceReader( BiopatMLFilePath ) );
FeatureList Matches = BioList[0].Search( SearchPosition, BioList[0].Length, MyPatterns.Pattern );
//According to Jacobi library total matches should be 57
Assert.AreEqual( 57, Matches.Count );
Assert.AreEqual( "Pribnow-box", Matches.Name );
//Perform some random checks from the 57 list
Match matched = (Match) Matches[10]; //try get the 11th matched
Assert.AreEqual( 0.66, matched.Similarity, 1e-2 );
Assert.AreEqual( 6, matched.Length );
Assert.AreEqual( "ttttat", matched.Letters() );
//try the first match
matched = (Match) Matches[0];
Assert.AreEqual( AlphabetFactory.Instance( AlphabetType.DNA ), matched.Alphabet );
Assert.AreEqual( 6, matched.Length );
Assert.AreEqual( 0.5, matched.Similarity, 1e-2 );
// Check the last match
matched = (Match) Matches[56];
Assert.AreEqual( 0.5, matched.Similarity, 1e-2 );
Assert.AreEqual( "tttctt", matched.Letters() );
}
public void TestStructuredPattern_SeriesAll()
{
BiopatMLFilePath = "BioPaperTestData/StructuredPattern/SeriesAll.xml";
using (BioPatMBF_Reader gbReader = new BioPatMBF_Reader())
{
BioList = gbReader.Read( Global.GetResourceReader( _singleDnaSeqGenBankFilename ) );
}
MyPatterns = DefinitionIO.Read( Global.GetResourceReader( BiopatMLFilePath ) );
FeatureList Matches = BioList[0].Search( SearchPosition, BioList[0].Length, MyPatterns.Pattern );
//expecting 49 matches based on the old jacobi result
Assert.AreEqual(49, Matches.Count);
Match matched = (Match)Matches[0]; //Query the 1st matched from the list of matches
Assert.AreEqual(3, matched.SubMatches.Count); //should have 3 sub matches
Assert.AreEqual("aattt", matched.SubMatches[0].Letters());
Assert.AreEqual("tataagtg", matched.SubMatches[1].Letters());
Assert.AreEqual("ttcaa", matched.SubMatches[2].Letters());
//And finally the main matched
Assert.AreEqual("aattttataagtgttcaa", matched.Letters());
}
/// <summary>
/// Constructs a Block of aligned sequences
/// (<see cref="QUT.Bio.BioPatML.Patterns.PWM"> PWM </see>).
/// </summary>
/// <param name="name">Name for element block</param>
/// <param name="sequenceList"> List of aligned sequences. </param>
/// <param name="background"> Histogram with base counts of the background
/// sequences.</param>
/// <param name="threshold"> Similarity threshold. </param>
public Block
(String name, SequenceList sequenceList,
HistogramSymbol background, double threshold)
: base(name, sequenceList[0].Alphabet, threshold)
{
Estimate(sequenceList, background);
}
/** 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);
}
public void TestRegionalPattern_Gap () {
BiopatMLFilePath = "BioPaperTestData/RegionalPattern/RegionalGap.xml";
BioPatMBF_Reader gbReader = new BioPatMBF_Reader();
BioList = gbReader.Read( Global.GetResourceReader( _singleProteinSeqGenBankFilename ) );
MyPatterns = DefinitionIO.Read( Global.GetResourceReader( BiopatMLFilePath ) );
FeatureList Matches = BioList[0].Search( SearchPosition, BioList[0].Length, MyPatterns.Pattern );
//Total matches according to old jacobi is 309
Assert.AreEqual( 410, Matches.Count );
//Checks the first match
Assert.AreEqual( "MT", Matches[0].MainSequence.Letters( Matches[0].Start, Matches[0].End ) );
//Checks if the last Match is in correect start and end pos
Assert.AreEqual( "CE", Matches[409].MainSequence.Letters( Matches[409].Start, Matches[409].End ) );
}
/// <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>
/// Estimates the weights of the PWM that's behind a Block pattern.
/// </summary>
/// <exception cref="System.ArgumentException">
/// Thrown when sequences length are not equal</exception>
/// <param name="sequenceList"> List of aligned sequences. </param>
/// <param name="background"> Histogram with base counts of the background
/// sequences. Can be null. In that case all frequencies are set equally.</param>
private void Estimate
(SequenceList sequenceList, HistogramSymbol background)
{
int length = sequenceList.MinLength();
if (sequenceList.MaxLength() != length)
throw new ArgumentException
("Sequences must be of equal length!");
if (background == null)
{
background = new HistogramSymbol();
foreach (Symbol sym in PWMalphabet)
background.Add(sym);
}
base.Init(length);
base.Estimate(sequenceList, 1, background);
}
public void TestRegionalPattern_Composition () {
BiopatMLFilePath = "BioPaperTestData/RegionalPattern/RegionalComposition.xml";
using ( BioPatMBF_Reader gbReader = new BioPatMBF_Reader() ) {
BioList = gbReader.Read( Global.GetResourceReader( _singleProteinSeqGenBankFilename ) );
}
MyPatterns = DefinitionIO.Read( Global.GetResourceReader( BiopatMLFilePath ) );
FeatureList Matches = BioList[0].Search( SearchPosition, BioList[0].Length, MyPatterns.Pattern );
//Total matches according to old jacobi is 309
Assert.AreEqual( 49, Matches.Count );
//Checks the first match
Assert.AreEqual( "GATLFKTRCLQCHTV", Matches[0].MainSequence.Letters( Matches[0].Start, Matches[0].End ) );
//Check see if the pattern used to match has the correct name for its matches
Assert.AreEqual( "transmembrane domain", Matches.Name );
Assert.AreEqual( 12, Matches[0].Start );
Assert.AreEqual( 26, Matches[0].End );
//Checks if the last Match is in correect start and end pos
Assert.AreEqual( "KDRNDLITYLKKACE", Matches[48].MainSequence.Letters( Matches[48].Start, Matches[48].End ) );
}
/// <summary>
/// Constructs a histogram based on the provided list of sequencs.
/// </summary>
/// <param name="sequenceList"> List of sequences. </param>
public void Add(SequenceList sequenceList)
{
for (int i = 0; i < sequenceList.Count; i++)
Add(sequenceList[i]);
}
/// <summary>
/// Constructs a histogram based on the provided list of sequencs.
/// </summary>
/// <param name="sequenceList"> List of sequences. </param>
public HistogramSymbol(SequenceList sequenceList)
{
Add(sequenceList);
}
/// <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);
}
public void TestMotifPattern_Block () {
BiopatMLFilePath = "BioPaperTestData/MotifPattern/Block.xml";
BioPatMBF_Reader gbReader = new BioPatMBF_Reader();
BioList = gbReader.Read( Global.GetResourceReader( _sampleGenBankFile2 ) );
MyPatterns = DefinitionIO.Read( Global.GetResourceReader( BiopatMLFilePath ) );
FeatureList Matches = BioList[0].Search( SearchPosition, BioList[0].Length, MyPatterns.Pattern );
Assert.AreEqual( 3, Matches.Count );
Assert.AreEqual( "Pribnow-box", Matches.Name );
Match matched = (Match) Matches[0];
Assert.AreEqual( 0.83, matched.Similarity, 1e-2 );
Assert.AreEqual( "tataac", matched.Letters() );
matched = (Match) Matches[1];
Assert.AreEqual( 0.76, matched.Similarity, 1e-2 );
Assert.AreEqual( "taacat", matched.Letters() );
matched = (Match) Matches[2];
Assert.AreEqual( 0.72, matched.Similarity, 1e-2 );
Assert.AreEqual( "cataaa", matched.Letters() );
}
public void TestMotifPattern_PWM () {
BiopatMLFilePath = "BioPaperTestData/MotifPattern/PWM.xml";
using ( BioPatMBF_Reader gbReader = new BioPatMBF_Reader() ) {
BioList = gbReader.Read( Global.GetResourceReader( _sampleGenBankFile2 ) );
}
MyPatterns = DefinitionIO.Read( Global.GetResourceReader( BiopatMLFilePath ) );
FeatureList Matches = BioList[0].Search( SearchPosition, BioList[0].Length, MyPatterns.Pattern );
Assert.AreEqual( 14, Matches.Count );
Assert.AreEqual( "Pribnow-box", Matches.Name );
Match matched = (Match) Matches[0];
Assert.AreEqual( 17, matched.Start );
Assert.AreEqual( 22, matched.End );
Assert.AreEqual( 0.67, matched.Similarity, 1e-2 );
Assert.AreEqual( "tctcct", matched.Letters() );
matched = (Match) Matches[1];
Assert.AreEqual( 25, matched.Start );
Assert.AreEqual( 30, matched.End );
Assert.AreEqual( 0.61, matched.Similarity, 1e-2 );
Assert.AreEqual( "ttggct", matched.Letters() );
matched = (Match) Matches[13];
Assert.AreEqual( 268, matched.Start );
Assert.AreEqual( 273, matched.End );
Assert.AreEqual( 0.67, matched.Similarity, 1e-2 );
Assert.AreEqual( "tgtgct", matched.Letters() );
}