Compilation and installation instructions added for Fedora

Bandage.pro

@@ -31,6 +31,7 @@ INCLUDEPATH += ui
 
 SOURCES += \
     program/main.cpp\
+    program/dotplot.cpp \
     program/settings.cpp \
     program/globals.cpp \
     program/graphlayoutworker.cpp \
@@ -121,6 +122,7 @@ SOURCES += \
 
 HEADERS  += \
     program/settings.h \
+    program/dotplot.h \
     program/globals.h \
     program/graphlayoutworker.h \
     graph/debruijnnode.h \
@@ -259,10 +261,6 @@ FORMS    += \
 RESOURCES += \
     images/images.qrc
 
-
-unix:INCLUDEPATH += /usr/include/
-unix:LIBS += -L/usr/lib
-
 # The following settings are compatible with OGDF being built in 64 bit release mode using Visual Studio 2013
 win32:LIBS += -lpsapi
 win32:RC_FILE = images/myapp.rc

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 \

README.md

@@ -37,16 +37,26 @@ The Linux binaries comes in two varieties: dynamically-linked and statically-lin
 
 If the compiled binaries do not work for you, the instructions below should help you build Bandage on most common OSs. If you are having difficulties building Bandage for your OS, feel free to contact me (Ryan) at [email protected] and I'll do my best to help you out!
 
-### <img src="http://rrwick.github.io/Bandage/images/OS/ubuntu.png" alt="" width="34" height="40" align="middle"> Ubuntu
+### <img src="http://rrwick.github.io/Bandage/images/OS/ubuntu.png" alt="" width="34" height="40" align="middle"> Ubuntu & Fedora
 
-The following instructions successfully build Bandage on a fresh installation of Ubuntu 14.04:
+The following instructions successfully build Bandage on a fresh installation of Ubuntu 14.04 or fedora 33:
 
 1. Ensure the package lists are up-to-date: `sudo apt-get update`
-2. Install prerequisite packages: `sudo apt-get install build-essential git qtbase5-dev libqt5svg5-dev`
+2. Install prerequisite packages: 
+
+    For Ubuntu:
+    ```
+    sudo apt-get install build-essential git qtbase5-dev libqt5svg5-dev
+    ```
+    For Fedora:
+    ```
+    sudo dnf groupinstall "Development Tools" "Development Libraries"
+    sudo dnf install qt5-qtsvg-devel git qt5-qtbase-devel
+    ```
 3. Download the Bandage code from GitHub: `git clone https://github.com/rrwick/Bandage.git`
 4. Open a terminal in the Bandage directory.
 5. Set the environment variable to specify that you will be using Qt 5, not Qt 4: `export QT_SELECT=5`
-6. Run qmake to generate a Makefile: `qmake`
+6. Run qmake to generate a Makefile: `qmake` or in Fedora `qmake-qt5`
 7. Build the program: `make`
 8. `Bandage` should now be an executable file.
 9. Optionally, copy the program into /usr/local/bin: `sudo make install`. The Bandage build directory can then be deleted.

blast/blastsearch.cpp

@@ -298,7 +298,7 @@ int BlastSearch::loadBlastQueriesFromFastaFile(QString fullFileName)
 
     std::vector<QString> queryNames;
     std::vector<QByteArray> querySequences;
-    AssemblyGraph::readFastaFile(fullFileName, &queryNames, &querySequences);
+    AssemblyGraph::readFastaOrFastqFile(fullFileName, &queryNames, &querySequences);
 
     for (size_t i = 0; i < queryNames.size(); ++i)
     {

command_line/image.cpp

@@ -130,6 +130,26 @@ int bandageImage(QStringList arguments)
                                                                                   g_settings->startingNodes,
                                                                                   "all");
 
+    QString errormsg;
+    QStringList columns;
+    bool coloursLoaded = false;
+    QString csvPath = parseColorsOption(arguments);
+    if (csvPath != "")
+    {
+        if(!g_assemblyGraph->loadCSV(csvPath, &columns, &errormsg, &coloursLoaded))
+        {
+            err << errormsg << endl;
+            return 1;
+        }
+
+        if(coloursLoaded == false)
+        {
+            err << csvPath << " didn't contains color" << endl;
+            return 1;
+        }
+         g_settings->nodeColourScheme = CUSTOM_COLOURS;
+    }
+
     if (errorMessage != "")
     {
         err << errorMessage << endl;
@@ -212,6 +232,7 @@ void printImageUsage(QTextStream * out, bool all)
     text << "";
     text << "Options:  --height <int>      Image height (default: 1000)";
     text << "--width <int>       Image width (default: not set)";
+    text << "--color <file>       csv file with 2 column first the node name second the node color";
     text << "";
     text << "If only height or width is set, the other will be determined automatically. If both are set, the image will be exactly that size.";
     text << "";
@@ -232,6 +253,9 @@ QString checkForInvalidImageOptions(QStringList arguments)
     error = checkOptionForInt("--width", &arguments, IntSetting(0, 1, 32767), false);
     if (error.length() > 0) return error;
 
+    error = checkOptionForString("--colors", &arguments, QStringList(), "a path of csv file");
+    if (error.length() > 0) return error;
+
     return checkForInvalidOrExcessSettings(&arguments);
 }
 
@@ -251,3 +275,15 @@ void parseImageOptions(QStringList arguments, int * width, int * height)
     parseSettings(arguments);
 }
 
+//This function parses the command line options. It assumes that the options
+//have already been checked for correctness.
+QString parseColorsOption(QStringList arguments)
+{
+    QString path = "";
+    if (isOptionPresent("--colors", &arguments))
+        path = getStringOption("--colors", &arguments);
+
+    parseSettings(arguments);
+
+    return path;
+}

command_line/image.h

@@ -28,5 +28,6 @@ int bandageImage(QStringList arguments);
 void printImageUsage(QTextStream * out, bool all);
 QString checkForInvalidImageOptions(QStringList arguments);
 void parseImageOptions(QStringList arguments, int * width, int * height);
+QString parseColorsOption(QStringList arguments);
 
 #endif // IMAGE_H

command_line/info.cpp

@@ -118,7 +118,7 @@ int bandageInfo(QStringList arguments)
         out << median << "\t";
         out << thirdQuartile << "\t";
         out << longestNode << "\t";
-        out << medianDepthByBase << "\n";
+        out << medianDepthByBase << "\t";
         out << estimatedSequenceLength << "\n";
     }
     else

graph/assemblygraph.cpp

@@ -616,7 +616,9 @@ void AssemblyGraph::buildDeBruijnGraphFromGfa(QString fullFileName, bool *unsupp
 
                 QString nodeName = lineParts.at(1);
                 if (nodeName.isEmpty())
-                    nodeName = "node";
+                    nodeName = getUniqueNodeName("node");
+                if (m_deBruijnGraphNodes.contains(nodeName + "+"))
+                    throw "load error";
 
                 QByteArray sequence = lineParts.at(2).toLocal8Bit();
 
@@ -972,6 +974,8 @@ void AssemblyGraph::buildDeBruijnGraphFromFastg(QString fullFileName)
                     nodeName += "-";
                 else
                     nodeName += "+";
+                if (m_deBruijnGraphNodes.contains(nodeName))
+                    throw "load error";
 
                 QString nodeDepthString = thisNodeDetails.at(5);
                 if (negativeNode)
@@ -1414,6 +1418,16 @@ void AssemblyGraph::buildDeBruijnGraphFromPlainFasta(QString fullFileName)
             m_depthTag = "KC";
         }
 
+        // Check to see if the name matches SKESA format, in which case we can get the depth and node number.
+        else if (thisNodeDetails.size() >= 3 && thisNodeDetails[0] == "Contig" && thisNodeDetails[1].toInt() > 0) {
+            name = thisNodeDetails[1];
+            bool ok;
+            double convertedDepth = thisNodeDetails[2].toDouble(&ok);
+            if (ok)
+                depth = convertedDepth;
+            m_depthTag = "KC";
+        }
+
         // If it doesn't match, then we will use the sequence name up to the first space.
         else {
             QStringList nameParts = name.split(" ");
@@ -1439,6 +1453,10 @@ void AssemblyGraph::buildDeBruijnGraphFromPlainFasta(QString fullFileName)
         if (lowerName.contains("circular=true"))
             circularNodeNames.push_back(name);
 
+        // SKESA circularity
+        if (thisNodeDetails.size() == 4 and thisNodeDetails[3] == "Circ")
+            circularNodeNames.push_back(name);
+
         if (name.length() < 1)
             throw "load error";
 
@@ -1692,6 +1710,12 @@ bool AssemblyGraph::loadCSV(QString filename, QStringList * columns, QString * e
 //If the node name it finds does not end in a '+' or '-', it will add '+'.
 QString AssemblyGraph::getNodeNameFromString(QString string)
 {
+    // First check for the most obvious case, where the string is already a node name.
+    if (m_deBruijnGraphNodes.contains(string))
+        return string;
+    if (m_deBruijnGraphNodes.contains(string + "+"))
+        return string + "+";
+
     QStringList parts = string.split("_");
     if (parts.size() == 0)
         return "";
@@ -2375,7 +2399,25 @@ QString AssemblyGraph::getOppositeNodeName(QString nodeName)
 }
 
 
-void AssemblyGraph::readFastaFile(QString filename, std::vector<QString> * names, std::vector<QByteArray> *sequences)
+void AssemblyGraph::readFastaOrFastqFile(QString filename, std::vector<QString> * names,
+                                         std::vector<QByteArray> * sequences) {
+    QChar firstChar = 0;
+    QFile inputFile(filename);
+    if (inputFile.open(QIODevice::ReadOnly)) {
+        QTextStream in(&inputFile);
+        QString firstLine = in.readLine();
+        firstChar = firstLine.at(0);
+        inputFile.close();
+    }
+    if (firstChar == '>')
+        readFastaFile(filename, names, sequences);
+    else if (firstChar == '@')
+        readFastqFile(filename, names, sequences);
+}
+
+
+
+void AssemblyGraph::readFastaFile(QString filename, std::vector<QString> * names, std::vector<QByteArray> * sequences)
 {
     QFile inputFile(filename);
     if (inputFile.open(QIODevice::ReadOnly))
@@ -2423,6 +2465,36 @@ void AssemblyGraph::readFastaFile(QString filename, std::vector<QString> * names
 }
 
 
+void AssemblyGraph::readFastqFile(QString filename, std::vector<QString> * names, std::vector<QByteArray> * sequences)
+{
+    QFile inputFile(filename);
+    if (inputFile.open(QIODevice::ReadOnly))
+    {
+        QTextStream in(&inputFile);
+        while (!in.atEnd())
+        {
+            QApplication::processEvents();
+
+            QString name = in.readLine().simplified();
+            QByteArray sequence = in.readLine().simplified().toLocal8Bit();
+            in.readLine();  // separator
+            in.readLine();  // qualities
+
+            if (name.length() == 0)
+                continue;
+            if (sequence.length() == 0)
+                continue;
+            if (name.at(0) != '@')
+                continue;
+            name.remove(0, 1); //Remove '@' from start
+            names->push_back(name);
+            sequences->push_back(sequence);
+        }
+        inputFile.close();
+    }
+}
+
+
 void AssemblyGraph::recalculateAllDepthsRelativeToDrawnMean()
 {
     double meanDrawnDepth = getMeanDepth(true);

graph/assemblygraph.h

@@ -128,8 +128,12 @@ class AssemblyGraph : public QObject
     void setAllEdgesExactOverlap(int overlap);
     void autoDetermineAllEdgesExactOverlap();
 
+    static void readFastaOrFastqFile(QString filename, std::vector<QString> * names,
+                                     std::vector<QByteArray> * sequences);
     static void readFastaFile(QString filename, std::vector<QString> * names,
                               std::vector<QByteArray> * sequences);
+    static void readFastqFile(QString filename, std::vector<QString> * names,
+                              std::vector<QByteArray> * sequences);
 
     int getDrawnNodeCount() const;
     void deleteNodes(std::vector<DeBruijnNode *> * nodes);