]> git.donarmstrong.com Git - mothur.git/blobdiff - fullmatrix.cpp
working on libshuff
[mothur.git] / fullmatrix.cpp
index 6151a65202901fa24ec6ff91166017a682483c62..6c09639fcbb4c48497ba2b17a61851ba2435c6af 100644 (file)
@@ -78,7 +78,6 @@ void FullMatrix::readSquareMatrix(ifstream& filehandle) {
                reading = new Progress("Reading matrix:    ", numSeqs * numSeqs);
                
                int count = 0;
-               float distance;
                
                string group, name;
                
@@ -92,9 +91,8 @@ void FullMatrix::readSquareMatrix(ifstream& filehandle) {
                        if(group == "not found") {      cout << "Error: Sequence '" << name << "' was not found in the group file, please correct." << endl; exit(1); }
                                
                        for(int j=0;j<numSeqs;j++){
-                               filehandle >> distance;
-                                       
-                               matrix[i][j] = distance;
+                               filehandle >> matrix[i][j];
+                               
                                count++;
                                reading->update(count);
                        }
@@ -161,7 +159,7 @@ void FullMatrix::sortGroups(int low, int high){
        
                int i = low;
                int j = high;
-               int y = 0;
+               float y = 0;
                string name;
                
                /* compare value */
@@ -235,7 +233,7 @@ void FullMatrix::printMatrix(ostream& out) {
                for (int i = 0; i < numSeqs; i++) {
                        out << "row " << i << " group = " << index[i].groupname << " name = " << index[i].seqName << endl;
                        for (int j = 0; j < numSeqs; j++) {
-                               out << matrix[i][j] << " ";
+                               //out << matrix[i][j] << " ";
                        }
                        out << endl;
                }
@@ -252,57 +250,105 @@ void FullMatrix::printMatrix(ostream& out) {
 }
 
 /**************************************************************************/
-void FullMatrix::getMinsForRowsVectors(){
+void FullMatrix::setBounds(){
        try{
                numGroups = globaldata->gGroupmap->namesOfGroups.size();
                
                //sort globaldata->gGroupmap.namesOfGroups so that it will match the matrix
                sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
                
+               //one for each comparision
+               //minsForRows.resize(numGroups*numGroups);
+               
                /*************************************************/
                //find where in matrix each group starts and stops
                /*************************************************/
-               vector<int> bounds;  //bounds[1] = starting row in matrix from group B, bounds[2] = starting row in matrix from group C, bounds[3] = no need to find upper bound of C because its numSeqs.
                bounds.resize(numGroups);
                
                bounds[0] = 0;
-               bounds[numGroups] = numSeqs-1;
+               bounds[numGroups] = numSeqs;
+
                //for each group find bounds of subgroup/comparison
                for (int i = 1; i < numGroups; i++) {
-                       getBounds(bounds[i], globaldata->gGroupmap->namesOfGroups[i]);
+                       getBounds(bounds[i], globaldata->gGroupmap->namesOfGroups[i-1]);
                }
                
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function getMinsForRowsVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the FullMatrix class function getMinsForRowsVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+
+}
+/**************************************************************************/
+vector<float> FullMatrix::getMins(int x) {
+       try{    
+               //clear out old data
+               minsForRows.clear();
+               
                /************************************************************/
-               //fill the minsForRows vectors for each group the user wants
+               //fill the minsForRows vector for the box the user wants
                /************************************************************/
-               int countx = bounds[1]; //where second group starts
-               int county = bounds[1]; 
+               int count = 0;
+               int lowBoundx = bounds[0]; //where first group starts
+               int lowBoundy = bounds[0]; 
+               int highBoundx = bounds[1]; //where second group starts
+               int highBoundy = bounds[1]; 
                
-               //go through the entire matrix
-               for (int x = 0; x < numSeqs; x++) {
-                       for (int y = 0; y < numSeqs; y++) {
-                               //if have not changed groups
-                               if ((x < countx) && (y < county)) {
-                                       
-                               }
+               int countx = 1;  //index in bound
+               int county = 1; //index in bound
+               
+               //find the bounds for the box the user wants
+               for (int i = 0; i < (numGroups * numGroups); i++) {
+               
+                       //are you at the box?
+                       if (count == x) { break; }
+                       else { count++; }
+                       
+                       //move to next box
+                       if (county < numGroups) {
+                               county++;
+                               highBoundy = bounds[county];
+                               lowBoundy = bounds[county-1];
+                       }else{ //you are moving to a new row of "boxes"
+                               county = 1;
+                               countx++;
+                               highBoundx = bounds[countx];
+                               lowBoundx = bounds[countx-1];
+                               highBoundy = bounds[county];
+                               lowBoundy = bounds[county-1];
                        }
                }
-                                       
                                
+               //each row in the box
+               for (int x = lowBoundx; x < highBoundx; x++) {
+                       float min4Row = 100000.0;
+                       //each entry in that row
+                       for (int y = lowBoundy; y < highBoundy; y++) {
+                               //if you are not on the diagonal and you are less than previous minimum
+                               if ((x != y) && (matrix[x][y] < min4Row)) {
+                                       min4Row = matrix[x][y];
+                               }
+                       }
+                       //save minimum value for that row in minsForRows vector of vectors
+                       minsForRows.push_back(min4Row);
+               }
                        
-       
+               return minsForRows;
        }
        catch(exception& e) {
-               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function getMinsForRowsVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function getMins. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
                exit(1);
        }
        catch(...) {
-               cout << "An unknown error has occurred in the FullMatrix class function getMinsForRowsVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               cout << "An unknown error has occurred in the FullMatrix class function getMins. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
                exit(1);
        }
-
 }
-
 /**************************************************************************/
 void FullMatrix::getBounds(int& higher, string group) {
        try{
@@ -311,7 +357,7 @@ void FullMatrix::getBounds(int& higher, string group) {
                //for each group find bounds of subgroup/comparison
                for (it = index.begin(); it != index.end(); it++) {
                        if (it->second.groupname == group) {
-                               if (gotLower != true) { gotLower = true; }
+                               gotLower = true; 
                        }else if ((gotLower == true) && (it->second.groupname != group)) {  higher = it->first; break; }
                }
        
@@ -327,3 +373,215 @@ void FullMatrix::getBounds(int& higher, string group) {
 
 }
 
+/**************************************************************************/
+//print out matrix
+void FullMatrix::printMinsForRows(ostream& out) {
+       try{
+               for (int j = 0; j < minsForRows.size(); j++) {
+                       out << minsForRows[j] << " ";
+               }
+               out << endl;
+
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function printMatrix. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the FullMatrix class function printMatrix. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+
+}
+/**************************************************************************/
+//shuffles the sequences in the 2 groups passed in.
+void FullMatrix::shuffle(int box){
+       try{
+               vector<int> rows2Swap;
+               vector<int> shuffled;
+               float y = 0;
+               string name = "";
+               
+               /****************************/
+               //find the box the user wants
+               /****************************/
+               int count = 0;
+               int lowBoundy = bounds[0]; //where first group starts
+               int highBoundy = bounds[1]; //where second group starts
+               int county = 1; //index in bound
+               
+               //find the bounds for the box the user wants
+               for (int i = 0; i < (numGroups * numGroups); i++) {
+               
+                       //are you at the box?
+                       if (count == box) { break; }
+                       else { count++; }
+                       
+                       //move to next box
+                       if (county < numGroups) {
+                               county++;
+                               highBoundy = bounds[county];
+                               lowBoundy = bounds[county-1];
+                       }else{ //you are moving to a new row of "boxes"
+                               county = 1;
+                               highBoundy = bounds[county];
+                               lowBoundy = bounds[county-1];
+                       }
+               }
+       
+               /************************/
+               //save its rows locations
+               /************************/
+               //go through the matrix map to find the rows from groups you want to randomize
+               for (int y = lowBoundy; y < highBoundy; y++) {
+                       rows2Swap.push_back(y);
+                       shuffled.push_back(y);
+               }
+               
+               //randomize rows to shuffle in shuffled
+               random_shuffle(shuffled.begin(), shuffled.end());
+               
+               /***************************************/
+               //swap rows and columns to randomize box
+               /***************************************/
+               for (int i = 0; i < shuffled.size(); i++) {
+                       //record the swaps you are making so you can undo them in restore function
+                       restoreIndex[i].a = shuffled[i];
+                       restoreIndex[i].b = rows2Swap[i];
+                       
+                       /* swap rows*/
+                       for (int h = 0; h < numSeqs; h++) {
+                               y = matrix[shuffled[i]][h];
+                               matrix[shuffled[i]][h] = matrix[rows2Swap[i]][h]; 
+                               matrix[rows2Swap[i]][h] = y;
+                       }
+                               
+                       /* swap columns */
+                       for (int b = 0; b < numSeqs; b++) {
+                               y = matrix[b][shuffled[i]];
+                               matrix[b][shuffled[i]] = matrix[b][rows2Swap[i]]; 
+                               matrix[b][rows2Swap[i]] = y;
+                       }
+                               
+                       //swap map elements
+                       name = index[shuffled[i]].seqName;
+                       index[shuffled[i]].seqName = index[rows2Swap[i]].seqName;
+                       index[rows2Swap[i]].seqName = name;
+               }
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function shuffle. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the FullMatrix class function shuffle. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+} 
+/**************************************************************************/
+//unshuffles the matrix.
+void FullMatrix::restore(){
+       try{
+               float y = 0;
+               string name = "";
+
+               //reverse iterate through swaps and undo them to restore original matrix and index map.
+               for(it2 = restoreIndex.rbegin(); it2 != restoreIndex.rend(); it2++) {
+                       /* swap rows */
+                       for (int h = 0; h < numSeqs; h++) {
+                               y = matrix[it2->second.a][h];
+                               matrix[it2->second.a][h] = matrix[it2->second.b][h]; 
+                               matrix[it2->second.b][h] = y;
+                       }
+                       
+                       /* swap columns */
+                       for (int b = 0; b < numSeqs; b++) {
+                               y = matrix[b][it2->second.a];
+                               matrix[b][it2->second.a] = matrix[b][it2->second.b]; 
+                               matrix[b][it2->second.b] = y;
+                       }
+                       
+                               
+                       //swap map elements
+                       name = index[it2->second.a].seqName;
+                       index[it2->second.a].seqName = index[it2->second.b].seqName;
+                       index[it2->second.b].seqName = name;
+               }
+
+               //clear restore for next shuffle
+               restoreIndex.clear();
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function restore. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the FullMatrix class function restore. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+}  
+/**************************************************************************/
+void FullMatrix::getDist(vector<float>& distances) {
+       try{
+               map<float, float> dist;  //holds the distances for the integral form
+               map<float, float>::iterator it;
+
+               /************************************************************/
+               //fill the minsForRows vectors for each group the user wants
+               /************************************************************/
+               int lowBoundx = bounds[0]; //where first group starts
+               int lowBoundy = bounds[0]; 
+               int highBoundx = bounds[1]; //where second group starts
+               int highBoundy = bounds[1]; 
+               
+               int countx = 1;  //index in bound
+               int county = 1; //index in bound
+               
+               //go through each "box" in the matrix
+               for (int i = 0; i < (numGroups * numGroups); i++) {
+                       //each row in the box
+                       for (int x = lowBoundx; x < highBoundx; x++) {
+                               float min4Row = 100000.0;
+                               //each entry in that row
+                               for (int y = lowBoundy; y < highBoundy; y++) {
+                                       //if you are not on the diagonal and you are less than previous minimum
+                                       if ((x != y) && (matrix[x][y] < min4Row)){
+                                               min4Row = matrix[x][y];
+                                       }
+                               }
+                               //save minimum value 
+                               dist[min4Row] = min4Row;
+                       }
+                       
+                       //****** reset bounds to process next "box" ********
+                       //if you still have more "boxes" in that row
+                       if (county < numGroups) {
+                               county++;
+                               highBoundy = bounds[county];
+                               lowBoundy = bounds[county-1];
+                       }else{ //you are moving to a new row of "boxes"
+                               county = 1;
+                               countx++;
+                               highBoundx = bounds[countx];
+                               lowBoundx = bounds[countx-1];
+                               highBoundy = bounds[county];
+                               lowBoundy = bounds[county-1];
+                       }
+               }
+
+               //store distances in users vector
+               for (it = dist.begin(); it != dist.end(); it++) {
+                       distances.push_back(it->first);
+               }
+               
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the FullMatrix class Function restore. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the FullMatrix class function restore. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+}
+