public List<Matched> PerformSearch(string BioPatMLContent, SequenceContract sequence)
{
List<Matched> matchedList = new List<Matched>();
StringReader sr = new StringReader(BioPatMLContent);
Definition MyPatterns = null;
using (BioPatMLPatternReader reader = new BioPatMLPatternReader())
{
MyPatterns = reader.ReadBioPatML(sr);
}
Sequence targetSequence = new Sequence(sequence.AlphabetName, sequence.Characters);
SequenceList matches = new FeatureList();
IPattern MatchingPattern = MyPatterns.MainPattern;
#region If bioPatML pattern is SeriesBest OR SetBest : Search by best
if (MatchingPattern is SeriesBest
|| MatchingPattern is SetBest)
{
Match match = targetSequence.SearchBest(0, 0, MatchingPattern);
matches.Add(match);
}
#endregion
#region Else, pattern is Motif, Any, Anchor, Prosite, RegularEx : Search by normal
else
//The rest
{
matches = targetSequence.Search(1, targetSequence.Length, MatchingPattern);
}
#endregion
for (int i = 0; i < matches.Count; i++)
{
Match matched = matches[i] as Match;
matchedList.Add(new Matched
(matched.Similarity,
matched.Start, matched.End,
matched.Length, matched.Letters(), sequence.Name));
}
return matchedList;
}
public SequenceContract ParseGenbankFile(string gbContent)
{
StringReader sr = new StringReader(gbContent);
SequenceContract sequence = new SequenceContract();
SequenceList bioSeqList = null; //BioPatML's data structure
using (ReaderBase reader = new BioPatMBF_Reader())
{
bioSeqList = reader.Read(sr);
}
//for now we only always use the first sequence SequenceName
sequence.Name = bioSeqList[0].Annotations("SequenceName").AnnotationValue as string;
//sequence.Name = bioSeqList[0].Name;
sequence.AlphabetName = bioSeqList[0].Alphabet.Name;
sequence.Characters = bioSeqList[0].Letters();
return sequence;
}