private IEnumerable <TagSequenceMatch> GetMatches(IEnumerable <SequenceTag.SequenceTag> tags, ProductSpectrum spec, IScorer scorer)
{
// Match tags against the database
var proteinsToTags = GetProteinToMatchedTagsMap(tags, _searchableDb, _aaSet, _tolerance, _tolerance);
//var tagSequenceMatchList = new List<TagSequenceMatch>();
// Extend matches
foreach (var entry in proteinsToTags)
{
var proteinName = entry.Key;
var matchedTagSet = entry.Value;
var proteinSequence = matchedTagSet.Sequence;
var tagFinder = new TagMatchFinder(spec, scorer, _featureFinder, proteinSequence, _tolerance, _aaSet, _maxSequenceMass);
foreach (var matchedTag in matchedTagSet.Tags)
{
if (matchedTag.Length < _minMatchedTagLength)
{
continue;
}
if (matchedTag.NTermFlankingMass == null && matchedTag.CTermFlankingMass == null)
{
continue;
}
var matches = tagFinder.FindMatches(matchedTag).ToArray();
//var prevScore = double.NegativeInfinity;
//foreach (var match in matches.OrderByDescending(m => m.Score))
foreach (var match in matches)
{
var sequence = proteinSequence.Substring(match.StartIndex, match.EndIndex - match.StartIndex);
//re-scoring
var sequenceObj = Sequence.CreateSequence(sequence, match.ModificationText, _aaSet);
match.Score = sequenceObj.GetInternalCleavages().Sum(c => scorer.GetFragmentScore(c.PrefixComposition, c.SuffixComposition));
//var numMatches = matchedTag.Length * 2 + match.NTermScore + match.CTermScore;
//var score = match.NTermScore + match.CTermScore;
//score += (matchedTag.NumReliableNTermFlankingMasses > 0)
// ? matchedTag.Length*CompositeScorer.ScoreParam.Prefix.ConsecutiveMatch
//: matchedTag.Length*CompositeScorer.ScoreParam.Suffix.ConsecutiveMatch;
// Poisson p-value score
//var n = (match.EndIndex - match.StartIndex - 1)*2;
//var lambda = numMatches / n;
//var pValue = 1 - Poisson.CDF(lambda, numMatches);
//var pScore = (pValue > 0) ? - Math.Log(pValue, 2) : 50.0;
//if (numMatches < 5) break;
//if (prevScore - numMatches > 2) break;
//prevScore = numMatches;
var pre = match.StartIndex == 0 ? '-' : proteinSequence[match.StartIndex - 1]; // startIndex is inclusive
var post = match.EndIndex >= proteinSequence.Length ? '-' : proteinSequence[match.EndIndex]; // endIndex is Exclusive
yield return(new TagSequenceMatch(sequence, proteinName, match, pre, post));
//tagSequenceMatchList.Add(new TagSequenceMatch(sequence, proteinName, match, pre, post));
}
}
}
//return tagSequenceMatchList;
}
private IEnumerable<TagSequenceMatch> GetMatches(IEnumerable<SequenceTag> tags, ProductSpectrum spec, IScorer scorer)
{
// Match tags against the database
var proteinsToTags = GetProteinToMatchedTagsMap(tags, _searchableDb, _aaSet, _tolerance, _tolerance);
//var tagSequenceMatchList = new List<TagSequenceMatch>();
// Extend matches
foreach (var entry in proteinsToTags)
{
var proteinName = entry.Key;
var matchedTagSet = entry.Value;
var proteinSequence = matchedTagSet.Sequence;
var tagFinder = new TagMatchFinder(spec, scorer, _featureFinder, proteinSequence, _tolerance, _aaSet, _maxSequenceMass);
foreach (var matchedTag in matchedTagSet.Tags)
{
if (matchedTag.Length < _minMatchedTagLength) continue;
if (matchedTag.NTermFlankingMass == null && matchedTag.CTermFlankingMass == null) continue;
var matches = tagFinder.FindMatches(matchedTag).ToArray();
//var prevScore = double.NegativeInfinity;
//foreach (var match in matches.OrderByDescending(m => m.Score))
foreach(var match in matches)
{
var sequence = proteinSequence.Substring(match.StartIndex, match.EndIndex - match.StartIndex);
//re-scoring
var sequenceObj = Sequence.CreateSequence(sequence, match.ModificationText, _aaSet);
match.Score = sequenceObj.GetInternalCleavages().Sum(c => scorer.GetFragmentScore(c.PrefixComposition, c.SuffixComposition));
//var numMatches = matchedTag.Length * 2 + match.NTermScore + match.CTermScore;
//var score = match.NTermScore + match.CTermScore;
//score += (matchedTag.NumReliableNTermFlankingMasses > 0)
// ? matchedTag.Length*CompositeScorer.ScoreParam.Prefix.ConsecutiveMatch
//: matchedTag.Length*CompositeScorer.ScoreParam.Suffix.ConsecutiveMatch;
// Poisson p-value score
//var n = (match.EndIndex - match.StartIndex - 1)*2;
//var lambda = numMatches / n;
//var pValue = 1 - Poisson.CDF(lambda, numMatches);
//var pScore = (pValue > 0) ? - Math.Log(pValue, 2) : 50.0;
//if (numMatches < 5) break;
//if (prevScore - numMatches > 2) break;
//prevScore = numMatches;
var pre = match.StartIndex == 0 ? '-' : proteinSequence[match.StartIndex - 1]; // startIndex is inclusive
var post = match.EndIndex >= proteinSequence.Length ? '-' : proteinSequence[match.EndIndex]; // endIndex is Exclusive
yield return new TagSequenceMatch(sequence, proteinName, match, pre, post);
//tagSequenceMatchList.Add(new TagSequenceMatch(sequence, proteinName, match, pre, post));
}
}
}
//return tagSequenceMatchList;
}