/// <summary>
/// Generates candidate matches between alignee features and baseline features.
/// It does so by finding all alignee-baseline feature pairs that match within a provided
/// mass tolerance window.
/// This method matches in mass only.
/// </summary>
/// <param name="aligneeFeatures"></param>
/// <param name="baselineFeatures"></param>
/// <param name="separationTypes">Separation types to include in matching.</param>
public void GenerateCandidateMatches(List <UMCLight> aligneeFeatures, List <UMCLight> baselineFeatures, IEnumerable <FeatureLight.SeparationTypes> separationTypes = null)
{
// Sort features by mass
var massComparer = new UMCLight.UmcMassComparer();
aligneeFeatures.Sort(massComparer);
baselineFeatures.Sort(massComparer);
// Go through each MassTimeFeature and see if the next baseline MassTimeFeature matches it
var baselineFeatureIndex = 0;
var featureMatches = new List <LcmsWarpFeatureMatch>();
foreach (var aligneeFeature in aligneeFeatures)
{
// Convert tolerance from ppm to Dalton
var massToleranceDa = aligneeFeature.MassMonoisotopic * this.options.MassTolerance / 1000000;
// Backtrack baselineFeatureIndex while the baseline feature's mass is greater than the candidate feature's mass minus massToleranceDa
while (baselineFeatureIndex == baselineFeatures.Count || baselineFeatureIndex >= 0 &&
(baselineFeatures[baselineFeatureIndex].MassMonoisotopic > aligneeFeature.MassMonoisotopic - massToleranceDa))
{
baselineFeatureIndex--;
}
baselineFeatureIndex++;
// Add candidate matches
while (baselineFeatureIndex < baselineFeatures.Count &&
(baselineFeatures[baselineFeatureIndex].MassMonoisotopic <
(aligneeFeature.MassMonoisotopic + massToleranceDa)))
{
var baselineFeature = baselineFeatures[baselineFeatureIndex];
if (baselineFeature.MassMonoisotopic >
(aligneeFeature.MassMonoisotopic - massToleranceDa))
{ // Feature is within mass tolerance, add the match.
var matchToAdd = new LcmsWarpFeatureMatch
{
AligneeFeature = aligneeFeature,
BaselineFeature = baselineFeature,
Net = aligneeFeature.Net,
BaselineNet = baselineFeature.Net
};
featureMatches.Add(matchToAdd);
}
baselineFeatureIndex++;
}
}
// Filter out ambiguous matches
featureMatches = this.RemovePromiscuousMatches(featureMatches);
this.Matches = featureMatches;
}
public List <LcmsWarpFeatureMatch> CalculateAlignmentMatches(List <UMCLight> aligneeFeatures, List <UMCLight> baselineFeatures, double netStdDev, double massStdDev)
{
// Sort features by mass
var massComparer = new UMCLight.UmcMassComparer();
aligneeFeatures.Sort(massComparer);
baselineFeatures.Sort(massComparer);
var baselineFeatureIndex = 0;
var featureMatches = new List <LcmsWarpFeatureMatch>();
var minMatchScore = -0.5 * (this.options.MassTolerance * this.options.MassTolerance) / (massStdDev * massStdDev);
minMatchScore -= 0.5 * (this.options.NetTolerance * this.options.NetTolerance) / (netStdDev * netStdDev);
foreach (var aligneeFeature in aligneeFeatures)
{
// Convert tolerance from ppm to Dalton
var massTolerance = aligneeFeature.MassMonoisotopic * this.options.MassTolerance / 1000000;
while (baselineFeatureIndex == baselineFeatures.Count || baselineFeatureIndex >= 0 &&
baselineFeatures[baselineFeatureIndex].MassMonoisotopic >
aligneeFeature.MassMonoisotopic - massTolerance)
{
baselineFeatureIndex--;
}
baselineFeatureIndex++;
LcmsWarpFeatureMatch bestMatchFeature = null;
var bestMatchScore = minMatchScore;
while (baselineFeatureIndex < baselineFeatures.Count &&
baselineFeatures[baselineFeatureIndex].MassMonoisotopic <
aligneeFeature.MassMonoisotopic + massTolerance)
{
var baselineFeature = baselineFeatures[baselineFeatureIndex];
if (baselineFeature.MassMonoisotopic >
aligneeFeature.MassMonoisotopic - massTolerance)
{
// Calculate the mass and net errors
// Compute as observedValue - expectedValue
var netDiff = aligneeFeature.NetAligned - baselineFeature.Net;
var driftDiff = aligneeFeature.DriftTime - baselineFeature.DriftTime;
var massDiff = aligneeFeature.MassMonoisotopic - baselineFeature.MassMonoisotopic;
var massDiffPpm = massDiff * 1000000.0 / baselineFeature.MassMonoisotopic;
var massDiffOriginal = aligneeFeature.MassMonoisotopicOriginal - baselineFeature.MassMonoisotopic;
var originalMassDiffPpm = massDiffOriginal * 1000000.0 / baselineFeature.MassMonoisotopic;
// Calculate the match score.
var matchScore = -0.5 * (netDiff * netDiff) / (netStdDev * netStdDev);
matchScore -= 0.5 * (massDiffPpm * massDiffPpm) / (massStdDev * massStdDev);
// If the match score is greater than the best match score, update the holding item.
if (matchScore > bestMatchScore)
{
bestMatchScore = matchScore;
bestMatchFeature = new LcmsWarpFeatureMatch
{
AligneeFeature = aligneeFeature,
BaselineFeature = baselineFeature,
BaselineFeatureIndex = baselineFeatureIndex,
Net = aligneeFeature.Net,
NetError = netDiff,
MassError = massDiff,
PpmMassError = massDiffPpm,
PpmMassErrorOriginal = originalMassDiffPpm,
DriftError = driftDiff,
BaselineNet = baselineFeatures[baselineFeatureIndex].Net
};
}
}
baselineFeatureIndex++;
}
// If we found a match, add it to the list of matches.
if (bestMatchFeature != null)
{
featureMatches.Add(bestMatchFeature);
}
}
return(featureMatches);
}