gfa paths

Bandage.pro

@@ -29,6 +29,7 @@ INCLUDEPATH += ui
 
 SOURCES += \
     program/main.cpp\
+    program/dotplot.cpp \
     program/settings.cpp \
     program/globals.cpp \
     program/graphlayoutworker.cpp \
@@ -119,6 +120,7 @@ SOURCES += \
 
 HEADERS  += \
     program/settings.h \
+    program/dotplot.h \
     program/globals.h \
     program/graphlayoutworker.h \
     graph/debruijnnode.h \

BandageTests.pro

@@ -33,6 +33,7 @@ SOURCES += \
     program/settings.cpp \
     program/globals.cpp \
     program/graphlayoutworker.cpp \
+		program/dotplot.cpp \
     graph/debruijnnode.cpp \
     graph/debruijnedge.cpp \
     graph/graphicsitemnode.cpp \
@@ -123,6 +124,7 @@ HEADERS  += \
     program/settings.h \
     program/globals.h \
     program/graphlayoutworker.h \
+		program/dotplot.h \
     graph/debruijnnode.h \
     graph/debruijnedge.h \
     graph/graphicsitemnode.h \

command_line/image.cpp

@@ -128,7 +128,7 @@ int bandageImage(QStringList arguments)
     std::vector<DeBruijnNode *> startingNodes = g_assemblyGraph->getStartingNodes(&errorTitle, &errorMessage,
                                                                                   g_settings->doubleMode,
                                                                                   g_settings->startingNodes,
-                                                                                  "all");
+                                                                                  "all", "");
 
     QString errormsg;
     QStringList columns;

command_line/reduce.cpp

@@ -105,7 +105,7 @@ int bandageReduce(QStringList arguments)
     std::vector<DeBruijnNode *> startingNodes = g_assemblyGraph->getStartingNodes(&errorTitle, &errorMessage,
                                                                                   g_settings->doubleMode,
                                                                                   g_settings->startingNodes,
-                                                                                  "all");
+                                                                                  "all", "");
 
     if (errorMessage != "")
     {

graph/assemblygraph.cpp

@@ -66,13 +66,26 @@ AssemblyGraph::~AssemblyGraph()
 
 void AssemblyGraph::cleanUp()
 {
-    QMapIterator<QString, DeBruijnNode*> i(m_deBruijnGraphNodes);
-    while (i.hasNext())
     {
-        i.next();
-        delete i.value();
+        QMapIterator<QString, Path*> i(m_deBruijnGraphPaths);
+        while (i.hasNext())
+        {
+            i.next();
+            delete i.value();
+        }
+        m_deBruijnGraphPaths.clear();
+    }
+
+
+    {
+        QMapIterator<QString, DeBruijnNode*> i(m_deBruijnGraphNodes);
+        while (i.hasNext())
+        {
+            i.next();
+            delete i.value();
+        }
+        m_deBruijnGraphNodes.clear();
     }
-    m_deBruijnGraphNodes.clear();
 
     QMapIterator<QPair<DeBruijnNode*, DeBruijnNode*>, DeBruijnEdge*> j(m_deBruijnGraphEdges);
     while (j.hasNext())
@@ -396,6 +409,7 @@ void AssemblyGraph::determineGraphInfo()
     m_edgeCount = edgeCount;
     m_totalLength = totalLength;
     m_meanDepth = getMeanDepth();
+    m_pathCount = m_deBruijnGraphPaths.size();
 
     std::sort(nodeDepths.begin(), nodeDepths.end());
 
@@ -579,6 +593,7 @@ void AssemblyGraph::buildDeBruijnGraphFromGfa(QString fullFileName, bool *unsupp
 
         QMap<QString, QColor> colours;
         QMap<QString, QString> labels;
+        QMap<QString, QString> paths;
 
         QTextStream in(&inputFile);
         while (!in.atEnd()) {
@@ -770,6 +785,14 @@ void AssemblyGraph::buildDeBruijnGraphFromGfa(QString fullFileName, bool *unsupp
                     *unsupportedCigar = true;
                 }
             }
+
+            // Load paths
+            else if (lineParts.at(0) == "P") {
+                if (lineParts.size() < 4)
+                    continue;
+
+                paths.insert(lineParts.at(1), lineParts.at(2));
+            }
         }
 
         //Pair up reverse complements, creating them if necessary.
@@ -804,6 +827,22 @@ void AssemblyGraph::buildDeBruijnGraphFromGfa(QString fullFileName, bool *unsupp
             int overlap = edgeOverlaps[i];
             createDeBruijnEdge(node1Name, node2Name, overlap, EXACT_OVERLAP);
         }
+
+        // Create all the paths.
+        QMapIterator<QString, QString> p(paths);
+        while (p.hasNext()) {
+            p.next();
+            QString pathName = p.key();
+
+            QString pathStringFailure;
+            Path pp = Path::makeFromString(p.value(), false, &pathStringFailure);
+
+            if (pp.isEmpty()) {
+                std::cout << pathStringFailure.toUtf8().constData() << std::endl;
+            } else {
+                m_deBruijnGraphPaths.insert(pathName, new Path(pp));
+            }
+        }
     }
 
     if (m_deBruijnGraphNodes.size() == 0)
@@ -2002,7 +2041,7 @@ void AssemblyGraph::addGraphicsItemsToScene(MyGraphicsScene * scene)
 
 
 std::vector<DeBruijnNode *> AssemblyGraph::getStartingNodes(QString * errorTitle, QString * errorMessage, bool doubleMode,
-                                                            QString nodesList, QString blastQueryName)
+                                                            QString nodesList, QString blastQueryName, QString pathName)
 {
     std::vector<DeBruijnNode *> startingNodes;
 
@@ -2062,6 +2101,16 @@ std::vector<DeBruijnNode *> AssemblyGraph::getStartingNodes(QString * errorTitle
         }
     }
 
+    else if (g_settings->graphScope == AROUND_PATHS)
+    {
+        if (m_deBruijnGraphPaths.count(pathName) == 0)
+        {
+            *errorTitle = "Invalid path";
+            *errorMessage = "No path with such name is loaded";
+            return startingNodes;
+        }
+    }
+
     g_settings->doubleMode = doubleMode;
     clearOgdfGraphAndResetNodes();
 
@@ -2072,6 +2121,12 @@ std::vector<DeBruijnNode *> AssemblyGraph::getStartingNodes(QString * errorTitle
     else if (g_settings->graphScope == DEPTH_RANGE)
         startingNodes = getNodesInDepthRange(g_settings->minDepthRange,
                                                  g_settings->maxDepthRange);
+    else if (g_settings->graphScope == AROUND_PATHS) {
+        QList<DeBruijnNode *> nodes = m_deBruijnGraphPaths[pathName]->getNodes();
+
+        for (QList<DeBruijnNode *>::iterator i = nodes.begin(); i != nodes.end(); ++i)
+            startingNodes.push_back(*i);
+    }
 
     return startingNodes;
 }
@@ -3539,7 +3594,7 @@ void AssemblyGraph::getGraphComponentCountAndLargestComponentSize(int * componen
 
     QSet<DeBruijnNode *> visitedNodes;
     QList< QList<DeBruijnNode *> > connectedComponents;
-    
+
     //Loop through all positive nodes.
     QMapIterator<QString, DeBruijnNode*> i(m_deBruijnGraphNodes);
     while (i.hasNext())
@@ -3548,12 +3603,12 @@ void AssemblyGraph::getGraphComponentCountAndLargestComponentSize(int * componen
         DeBruijnNode * v = i.value();
         if (v->isNegativeNode())
             continue;
-        
+
         //If the node has not yet been visited, then it must be the start of a new connected component.
         if (!visitedNodes.contains(v))
         {
             QList<DeBruijnNode *> connectedComponent;
-            
+
             QQueue<DeBruijnNode *> q;
             q.enqueue(v);
             visitedNodes.insert(v);
@@ -3576,9 +3631,9 @@ void AssemblyGraph::getGraphComponentCountAndLargestComponentSize(int * componen
             }
 
             connectedComponents.push_back(connectedComponent);
-        }  
+        }
     }
-    
+
     //Now that the list of connected components is built, we look for the
     //largest one (as measured by total node length).
     *componentCount = connectedComponents.size();

graph/assemblygraph.h

@@ -50,13 +50,16 @@ class AssemblyGraph : public QObject
     //pointers.
     QMap<QPair<DeBruijnNode*, DeBruijnNode*>, DeBruijnEdge*> m_deBruijnGraphEdges;
 
+    QMap<QString, Path*> m_deBruijnGraphPaths;
+
     ogdf::Graph * m_ogdfGraph;
     ogdf::EdgeArray<double> * m_edgeArray;
     ogdf::GraphAttributes * m_graphAttributes;
 
     int m_kmer;
     int m_nodeCount;
     int m_edgeCount;
+    int m_pathCount;
     long long m_totalLength;
     long long m_shortestContig;
     long long m_longestContig;
@@ -115,7 +118,8 @@ class AssemblyGraph : public QObject
                                                  QString * errorMessage,
                                                  bool doubleMode,
                                                  QString nodesList,
-                                                 QString blastQueryName);
+                                                 QString blastQueryName,
+                                                 QString pathName);
 
     bool checkIfStringHasNodes(QString nodesString);
     QString generateNodesNotFoundErrorMessage(std::vector<QString> nodesNotInGraph,