Fixed centroiding algorithm to process peaks beyond 2000 m/z.

Added new parameter molecule_max_mz, which sets the upper boundary of molecules to find features for.

CHardklor2.cpp

@@ -336,6 +336,13 @@ void CHardklor2::Centroid(Spectrum& s, Spectrum& out){
 
 	out.clear();
 
+  //Get boundaries of the spectrum. centroids must be within boundaries.
+  double minMZ, maxMZ;
+  if(s.size()>0){
+    minMZ = s[0].mz;
+    maxMZ = s[s.size()-1].mz;
+  }
+
   bLastPos=false;
 	for(i=0;i<s.size()-1;i++){
 
@@ -385,9 +392,9 @@ void CHardklor2::Centroid(Spectrum& s, Spectrum& out){
 					centroid.intensity=s[bestPeak].intensity;
 				}
 
-				//Hack until I put in mass ranges
-				if(centroid.mz<0 || centroid.mz>2000) {
-					//do nothing if invalid mz
+				//Centroided peaks must fall within spectrum mass range
+				if(centroid.mz<minMZ || centroid.mz>maxMZ) {
+					//do nothing if invalid mz, but perhaps find a better way to handle this one day.
 				} else {
 					out.add(centroid);
 				}

CHardklorParser.cpp

@@ -242,6 +242,10 @@ void CHardklorParser::parse(char* cmd) {
     strcpy(global.MercuryFile,tok);
 
 	} else if(strcmp(param,"max_features")==0){
+
+  } else if(strcmp(param,"molecule_max_mz")==0){
+    global.maxMolMZ=atof(tok);
+
 	} else if(strcmp(param,"ms_level")==0){
     if(atoi(tok)==3){
       global.mzXMLFilter=MS3;

CHardklorSetting.cpp

@@ -73,6 +73,8 @@ CHardklorSetting::CHardklorSetting(){
   //rawAvgWidth=1;
   //rawAvgCutoff=1000;
 
+  maxMolMZ=5000;
+
   strcpy(rawFilter,"");
 }
 
@@ -134,6 +136,9 @@ CHardklorSetting::CHardklorSetting(const CHardklorSetting& c){
   //rawAvg=c.rawAvg;
   //rawAvgWidth=c.rawAvgWidth;
   //rawAvgCutoff=c.rawAvgCutoff;
+
+  maxMolMZ=c.maxMolMZ;
+
   strcpy(rawFilter,c.rawFilter);
 }
 
@@ -199,6 +204,7 @@ CHardklorSetting& CHardklorSetting::operator=(const CHardklorSetting& c){
     //rawAvg=c.rawAvg;
     //rawAvgWidth=c.rawAvgWidth;
     //rawAvgCutoff=c.rawAvgCutoff;
+    maxMolMZ = c.maxMolMZ;
 
     strcpy(rawFilter,c.rawFilter);
   }

CHardklorSetting.h

@@ -71,6 +71,7 @@ class CHardklorSetting {
   int smooth;       //Savitsky-Golay smoothing window size
   //int sna;          //Signal-to-noise algorithm; 0=THRASH, 1=Persistent peaks (PP)
 
+  double maxMolMZ;  //Largest m/z to detect in the data
   double corr;      //correlation threshold
   double ppm;       //ppm tolerance of m/z values to match across scans
   double res400;    //resolution at m/z 400

CModelLibrary.cpp

@@ -55,11 +55,11 @@ bool CModelLibrary::buildLibrary(int lowCharge, int highCharge, vector<CHardklor
 		return false;
 	}
 
-	//Fill in boundaries
+	//Fill in boundaries; Note Nov11_2019: This needs dynamic updating, perhaps drawn from a new parameter.
 	chargeMin=lowCharge;
 	chargeCount=highCharge+1;
 	varCount=(int)pepVariants.size();
-	merCount=1000;
+	if(merCount==0) merCount=1000;
 
 	libModel = new mercuryModel**[chargeCount];
 	for(i=chargeMin;i<chargeCount;i++){
@@ -127,12 +127,23 @@ void CModelLibrary::eraseLibrary(){
 	delete [] libModel;
 
 	libModel=NULL;
+  merCount=0;
 
 }
 
 mercuryModel* CModelLibrary::getModel(int charge, int var, double mz){
 
 	int intMZ=(int)(mz/5);
+
+  //punting this for now. A better solution needs to be made, but will wait until necessary.
+  if(intMZ>=merCount) {
+    cout << "Spectrum feature beyond molecule_max_mz. Hard boundary encountered. Please increase molecule_max_mz." << endl;
+    exit(-47);
+  }
 	return &libModel[charge][var][intMZ];
 
+}
+
+void CModelLibrary::setSize(double mz){
+  merCount=(int)(mz/5)+1;
 }

CModelLibrary.h

@@ -34,6 +34,7 @@ class CModelLibrary {
   bool buildLibrary(int lowCharge, int highCharge, std::vector<CHardklorVariant>& pepVariants);
 	void eraseLibrary();
 	mercuryModel* getModel(int charge, int var, double mz);
+  void setSize(double mz);
 
 protected:
 

HardklorApp.cpp

@@ -36,8 +36,8 @@ int main(int argc, char* argv[]) {
 	CMercury8 *mercury;
 	CModelLibrary *models;
 
-	cout << "Hardklor v2.3.1, April 23 2018" << endl;
-	cout << "Mike Hoopmann, Mike MacCoss\nCopyright 2007-2018\nUniversity of Washington" << endl;
+	cout << "Hardklor v2.3.2, November 11 2019" << endl;
+	cout << "Mike Hoopmann, Mike MacCoss\nCopyright 2007-2019\nUniversity of Washington" << endl;
 	if(argc < 2 || (argc>2 && strcmp(argv[1],"-cmd")!=0) ){
 		cout << "Usage:\t\thardklor <config file>\n";
     cout << "\t\thardklor -cmd [options] <input file> <output file>\n" << endl;
@@ -78,6 +78,7 @@ int main(int argc, char* argv[]) {
 			for(j=0;j<hp.queue(i).variant->size();j++)  pepVariants.push_back(hp.queue(i).variant->at(j));
 
 			models->eraseLibrary();
+      models->setSize(hp.queue(i).maxMolMZ);
 			models->buildLibrary(hp.queue(i).minCharge,hp.queue(i).maxCharge,pepVariants);
 			h2.GoHardklor(hp.queue(i));
 		} else {