public FeatureDetectionMethods(ThermoRawFile rawFile)
{
this.finalGroups = new List <FeatureGroup>();
this.eiSpectra = new List <EISpectrum>();
this.rawFile = rawFile;
this.norm = new Normalization();
}
public void PushFeaturesToDatabase(ThermoRawFile rawFile, List <Feature> allFeatures, SQLiteConnection conn)
{
AddRawFileEntry(rawFile, conn);
using (var transaction = conn.BeginTransaction())
{
if (!ContainsRawFileFeatures(rawFile, conn))
{
int id = GetRawFileID(rawFile, conn);
int count = 0;
int onePercentCount = (int)(0.01 * allFeatures.Count);
onePercentCount++;
foreach (var feature in allFeatures)
{
if (!ContainsSpecificFeature(feature.ID_Number, conn))
{
double apexIntensity = 0;
double apexRT = 0;
var smoothLine = "";
var rawLine = "";
foreach (var peak in feature.smoothRTPeaks)
{
if (peak.Intensity > apexIntensity)
{
apexIntensity = peak.Intensity;
apexRT = peak.RT;
}
smoothLine += peak.RT + "," + peak.Intensity + ";";
}
foreach (var peak in feature.RawRTPeaks)
{
rawLine += peak.RT + "," + peak.Intensity + ";";
}
var insertText = "INSERT INTO featureTable (ID, Name, mz, ApexRT, ApexIntensity, ID_Number, SmoothFeatureString, RawFeatureString)"
+
" VALUES (@ID, @Name, @mz, @ApexRT, @ApexIntensity, @ID_Number, @SmoothFeatureString, @RawFeatureString)";
var insertCommand = new SQLiteCommand(insertText, conn);
insertCommand.Parameters.AddWithValue("@ID", id);
insertCommand.Parameters.AddWithValue("@Name", rawFile.Name);
insertCommand.Parameters.AddWithValue("@mz", feature.averageMZ);
insertCommand.Parameters.AddWithValue("@ApexRT", apexRT);
insertCommand.Parameters.AddWithValue("@ApexIntensity", apexIntensity);
insertCommand.Parameters.AddWithValue("@SmoothFeatureString", smoothLine);
insertCommand.Parameters.AddWithValue("@RawFeatureString", rawLine);
insertCommand.Parameters.AddWithValue("@ID_Number", feature.ID_Number);
insertCommand.ExecuteNonQuery();
count++;
if (count % onePercentCount == 0)
{
OnProgressUpdate(.01);
}
}
}
}
transaction.Commit();
transaction.Dispose();
OnFinish();
}
}
public bool ContainsRawFileFeatures(ThermoRawFile rawFile, SQLiteConnection conn)
{
var queryText = "SELECT s.Name FROM featureTable s WHERE s.Name=@Name";
var queryCommand = new SQLiteCommand(queryText, conn);
queryCommand.Parameters.AddWithValue("@Name", rawFile.Name);
var reader = queryCommand.ExecuteReader();
return(reader.HasRows);
}
private void GroupFeatures()
{
if (currIndex < allRawFileNames.Count)
{
var currRawFileString = allRawFileNames[currIndex];
currRawFile = new ThermoRawFile(currRawFileString);
currRawFile.Open();
var path = outputTextBox.Text + "\\" + currRawFile.Name + "_ExtractedFeatures.gcfeat";
conn = new SQLiteConnection(@"Data Source=" + path);
conn.Open();
rawFileListBox.ClearSelected();
rawFileListBox.SelectedIndex = currIndex;
if (!SQLiteIOMethods.IsGroupingDone(conn))
{
currFDM = new FeatureDetectionMethods(currRawFile);
statusLabel.Text = "Grouping Features from " + currRawFile.Name + "...";
currFDM.Finished += grouping_Finished;
//currentThread = new Thread(() => currFDM.GroupFeatures(conn));
//currentThread.Start();
}
else
{
currIndex++;
double percent = ((double)currIndex) / ((double)allRawFileNames.Count) * 100;
statusBar.Value = (int)percent;
GroupFeatures();
}
}
else
{
currRawFile.ClearCachedScans();
currRawFile.Dispose();
if (conn != null)
{
conn.Close();
conn.Dispose();
}
conn = null;
if (currFDM != null)
{
currFDM.Dispose();
currFDM = null;
}
GC.Collect();
rawFileListBox.ClearSelected();
foreach (Control control in this.Controls)
{
control.Enabled = true;
}
statusBar.Value = 100;
statusLabel.Text = "Done!";
}
}
public static bool ContainsRawFileFeatures(SQLiteConnection conn, ThermoRawFile raw)
{
var queryText = "SELECT s.ID FROM featureTable s WHERE s.Name=@Name";
var queryCommand = new SQLiteCommand(queryText, conn);
queryCommand.Parameters.AddWithValue("@Name", raw.Name);
var reader = queryCommand.ExecuteReader();
var hasRow = reader.HasRows;
reader.Close();
reader.Dispose();
return(hasRow);
}
public static int GetRawFileID(ThermoRawFile raw, SQLiteConnection conn)
{
var queryText = "SELECT s.ID FROM rawFileNameTable s WHERE s.Name=@Name";
var queryCommand = new SQLiteCommand(queryText, conn);
queryCommand.Parameters.AddWithValue("@Name", raw.Name);
var reader = queryCommand.ExecuteReader();
int id = 0;
while (reader.Read())
{
id = int.Parse(reader["ID"].ToString());
}
return(id);
}
public static void AddRawFileTICChroma(ThermoRawFile rawFile, SQLiteConnection conn)
{
if (!ContainsRawFileTIC(conn, rawFile))
{
var ticChroma = rawFile.GetTICChroma();
var line = "";
foreach (var peak in ticChroma)
{
line += peak.Time + "," + peak.Intensity + ";";
}
int id = GetRawFileID(rawFile, conn);
var insertText = "INSERT INTO ticTable (ID, Name, TICChroma) VALUES (@ID, @Name, @TICChroma)";
var insertCommand = new SQLiteCommand(insertText, conn);
insertCommand.Parameters.AddWithValue("@ID", id);
insertCommand.Parameters.AddWithValue("@Name", rawFile.Name);
insertCommand.Parameters.AddWithValue("@TICChroma", line);
insertCommand.ExecuteNonQuery();
}
}
public void PrepDatabase(string rawFilePath, string outputDirectory)
{
var rawFile = new ThermoRawFile(rawFilePath);
rawFile.Open();
var path = outputDirectory + "\\" + rawFile.Name + "_ExtractedFeatures.gcfeat";
var conn = new SQLiteConnection(@"Data Source=" + path);
conn.Open();
SQLiteIOMethods.CreateTablesInDatabase(conn);
SQLiteIOMethods.AddRawFileEntry(rawFile, conn);
SQLiteIOMethods.AddRawFileTICChroma(rawFile, conn);
SQLiteIOMethods.CreateIndices(conn);
conn.Close();
conn.Dispose();
rawFile.ClearCachedScans();
rawFile.Dispose();
OnFinish();
}
public static void AddRawFileEntry(ThermoRawFile rawFile, SQLiteConnection conn)
{
List <int> ids = new List <int>();
var queryCommandText = "SELECT s.ID FROM rawFileNameTable s";
var queryCommand = new SQLiteCommand(queryCommandText, conn);
var reader = queryCommand.ExecuteReader();
while (reader.Read())
{
ids.Add(int.Parse(reader["ID"].ToString()));
}
ids = ids.OrderBy(x => x).ToList();
if (ids.Count == 0)
{
int id = 1;
queryCommandText = "INSERT INTO rawFileNameTable (ID, Name, Extracted, Grouped) VALUES (@ID, @Name, @Extracted, @Grouped)";
queryCommand = new SQLiteCommand(queryCommandText, conn);
queryCommand.Parameters.AddWithValue("@ID", id);
queryCommand.Parameters.AddWithValue("@Name", rawFile.Name);
queryCommand.Parameters.AddWithValue("@Extracted", false);
queryCommand.Parameters.AddWithValue("@Grouped", false);
queryCommand.ExecuteNonQuery();
queryCommand.Dispose();
}
else
{
int nextID = ids.Last() + 1;
if (!ContainsRawFile(conn, rawFile))
{
var insertText = "INSERT INTO rawFileNameTable (ID, Name, Extracted) VALUES (@ID, @Name, @Extracted, @Grouped)";
var insertCommand = new SQLiteCommand(insertText, conn);
insertCommand.Parameters.AddWithValue("@ID", nextID);
insertCommand.Parameters.AddWithValue("@Name", rawFile.Name);
insertCommand.Parameters.AddWithValue("@Extracted", false);
insertCommand.Parameters.AddWithValue("@Grouped", false);
insertCommand.ExecuteNonQuery();
nextID++;
}
}
}
public List <MZPeak> GetPeaksFromNearestSpectrum(FeatureGroup group, ThermoRawFile rawFile)
{
int scanNum = rawFile.GetSpectrumNumber(group.ApexTime);
List <MZPeak> returnPeaks = new List <MZPeak>();
ThermoSpectrum spectrum = rawFile.GetLabeledSpectrum(scanNum);
HashSet <double> peakMZs = new HashSet <double>();
foreach (Feature feature in group.allFeatures)
{
MZPeak outPeak = spectrum.GetClosestPeak(MassRange.FromPPM(feature.AverageMZ, 10));
if (outPeak != null)
{
if (!peakMZs.Contains(outPeak.Intensity))
{
returnPeaks.Add(new MZPeak(feature.AverageMZ, outPeak.Intensity));
peakMZs.Add(outPeak.Intensity);
}
}
}
returnPeaks = returnPeaks.OrderBy(x => x.MZ).ToList();
return(returnPeaks);
}
private void extractFeatures()
{
if (currIndex < allRawFileNames.Count)
{
var currRawFileString = allRawFileNames[currIndex];
currRawFile = new ThermoRawFile(currRawFileString);
currRawFile.Open();
var path = outputTextBox.Text + "\\" + currRawFile.Name + "_ExtractedFeatures.gcfeat";
conn = new SQLiteConnection(@"Data Source=" + path);
conn.Open();
rawFileListBox.ClearSelected();
rawFileListBox.SelectedIndex = currIndex;
this.Text = currIndex + "\\" + allRawFileNames.Count + " Deconvolved...";
if (!SQLiteIOMethods.IsExtractionDone(conn))
{
currFDM = new FeatureDetectionMethods(currRawFile);
statusLabel.Text = "Adding Features from " + currRawFile.Name + " to Database...";
currFDM.Progress += deconvolution_Progress;
currFDM.Finished += deconvolution_Finished;
currentThread = new Thread(() => currFDM.StepwiseSetupFinalGroups(currRawFile, conn));
currentThread.Start();
}
else
{
conn.Close();
conn.Dispose();
currRawFile.Dispose();
currIndex++;
extractFeatures();
}
}
else
{
currIndex = 0;
//statusBar.Value = 0;
//statusLabel.Text = "Starting Feature Grouping...";
//GroupFeatures();
currRawFile.ClearCachedScans();
currRawFile.Dispose();
if (conn != null)
{
conn.Close();
conn.Dispose();
}
conn = null;
if (currFDM != null)
{
currFDM.Dispose();
currFDM = null;
}
GC.Collect();
rawFileListBox.ClearSelected();
foreach (Control control in this.Controls)
{
control.Enabled = true;
}
statusBar.Value = 100;
statusLabel.Text = "Done!";
this.Text = "Deconvolution Engine";
}
}
public void StepwiseSetupFinalGroups(ThermoRawFile rawfile, SQLiteConnection conn)
{
List <int> list = new List <int>();
foreach (var scan in rawFile.GetMsScans())
{
if (scan.MsnOrder == 1)
{
list.Add(scan.SpectrumNumber);
}
}
var extractedFeatures = ExtractFeatures(list, 10).ToList();
//Features have been extracted
//Check which features are valid and which are not based on how their max intensity changes
//nwk temp code
var tmpFeatures = extractedFeatures.Where(x => x.ApexTime > 13.55 && x.apexTime <13.70 && x.AverageMZ> 267.082 && x.averageMZ < 267.095).ToList();
var q = "";
CheckPeakRise(extractedFeatures);
//Smooth features now
foreach (Feature feature in extractedFeatures)
{
feature.SmoothRTPeaks = GetRollingAveragePeaks(feature.RawRTPeaks, 9);
}
CheckPeakRiseSmooth(extractedFeatures);
List <Feature> finalFeatures = GetFinalFeatures(extractedFeatures);
extractedFeatures.Clear();
CheckPeakRiseSmooth(finalFeatures);
int count = 1;
foreach (var feat in finalFeatures)
{
feat.ID_Number = count;
count++;
}
//push features to database
PushFeaturesToDatabase(rawfile, finalFeatures, conn);
List <FeatureGroup> groups = GroupFeatures(finalFeatures, .04, 2).ToList();
DoubleBack(groups, finalFeatures);
finalFeatures.Clear();
List <FeatureGroup> allSubGroups = new List <FeatureGroup>();
foreach (FeatureGroup group in groups)
{
GetMainSubgroup(group);
GetFeatureSubGroups(group);
if (group.mainSubGroup.allFeatures.Count > 1)
{
allSubGroups.Add(group.mainSubGroup);
}
foreach (FeatureGroup subGroup in group.SubGroups)
{
allSubGroups.Add(subGroup);
}
}
this.finalGroups = GetFinalFeatures(allSubGroups);
foreach (FeatureGroup group in this.finalGroups)
{
group.finalPeaks = GetPeaksFromNearestSpectrum(group, rawFile);
}
foreach (FeatureGroup group in this.finalGroups)
{
EISpectrum spec = new EISpectrum();
spec.ApexTimeEI = group.ApexTime;
spec.FeatureGroup = group;
spec.FinalEIPeaks = group.finalPeaks;
spec.FinalEIPeaks = spec.FinalEIPeaks.OrderByDescending(x => x.Intensity).ToList();
if (group.finalPeaks.Count >= 2) //here as well (changed to remove num feature restriction)
{
spec.BasePeak = spec.FinalEIPeaks[0];
this.eiSpectra.Add(spec);
}
spec.FinalNormalizedEIPeaks.AddRange(spec.FinalEIPeaks);
spec.FinalNormalizedEIPeaks = spec.FinalNormalizedEIPeaks.OrderBy(x => x.MZ).ToList();
norm.CombineLikeMZPeaks(spec.FinalNormalizedEIPeaks);
spec.AdjustedPeaks.AddRange(spec.FinalNormalizedEIPeaks);
norm.GetAdjustedPeaks(spec.AdjustedPeaks);
spec.DenominatorTerm = norm.GetDemoninatorTerm(spec.AdjustedPeaks);
}
//this.eiSpectra = CleanUpSpectra(eiSpectra);
this.eiSpectra = eiSpectra.Where(x => x.FinalEIPeaks.Count >= 2).ToList();
AddFeatureGroupsToDatabase(this.eiSpectra, conn);
SQLiteIOMethods.ExtractionComplete(conn);
SQLiteIOMethods.GroupingComplete(conn);
OnFinish();
}
private void StepwiseSetupFinalGroups(ThermoRawFile rawFile)
{
List <int> list = new List <int>();
foreach (var scan in rawFile.GetMsScans())
{
if (scan.RetentionTime < 7)
{
list.Add(scan.SpectrumNumber);
}
}
var extractedFeatures = ExtractFeatures(list, 10).ToList();
//Features have been extracted
//Check which features are valid and which are not based on how their max intensity changes
CheckPeakRise(extractedFeatures);
//Smooth features now
foreach (Feature feature in extractedFeatures)
{
feature.SmoothRTPeaks = GetRollingAveragePeaks(feature.RawRTPeaks, 9);
}
CheckPeakRiseSmooth(extractedFeatures);
List <Feature> finalFeatures = GetFinalFeatures(extractedFeatures);
extractedFeatures.Clear();
CheckPeakRiseSmooth(finalFeatures);
List <FeatureGroup> groups = GroupFeatures(finalFeatures, .04, 2).ToList();
DoubleBack(groups, finalFeatures);
finalFeatures.Clear();
List <FeatureGroup> allSubGroups = new List <FeatureGroup>();
foreach (FeatureGroup group in groups)
{
GetMainSubgroup(group);
GetFeatureSubGroups(group);
if (group.mainSubGroup.allFeatures.Count > 1)
{
allSubGroups.Add(group.mainSubGroup);
}
foreach (FeatureGroup subGroup in group.SubGroups)
{
allSubGroups.Add(subGroup);
}
}
this.finalGroups = GetFinalFeatures(allSubGroups);
foreach (FeatureGroup group in this.finalGroups)
{
group.finalPeaks = GetPeaksFromNearestSpectrum(group, rawFile);
}
foreach (FeatureGroup group in this.finalGroups)
{
EISpectrum spec = new EISpectrum();
spec.ApexTimeEI = group.ApexTime;
spec.FeatureGroup = group;
spec.FinalEIPeaks = group.finalPeaks;
spec.FinalEIPeaks = spec.FinalEIPeaks.OrderByDescending(x => x.Intensity).ToList();
if (group.finalPeaks.Count > 6) //here as well (changed to remove num feature restriction)
{
spec.BasePeak = spec.FinalEIPeaks[0];
this.eiSpectra.Add(spec);
}
spec.FinalNormalizedEIPeaks.AddRange(spec.FinalEIPeaks);
spec.FinalNormalizedEIPeaks = spec.FinalNormalizedEIPeaks.OrderBy(x => x.MZ).ToList();
norm.CombineLikeMZPeaks(spec.FinalNormalizedEIPeaks);
spec.AdjustedPeaks.AddRange(spec.FinalNormalizedEIPeaks);
norm.GetAdjustedPeaks(spec.AdjustedPeaks);
spec.DenominatorTerm = norm.GetDemoninatorTerm(spec.AdjustedPeaks);
}
OnFinish();
}
