Preparing release v0.4.1

HisMaker.hh

@@ -4,7 +4,11 @@
 // C/C++ includes
 #include <iostream> 
 #include <fstream> 
-#include <sstream> 
+#include <sstream>
+#include <string>
+#include <map>
+#include <vector>
+
 using namespace std; 
 
 // ROOT includes

INSTALL

@@ -0,0 +1,138 @@
+Building, Installing and Running CNVnator
+=========================================
+
+Requirements
+============
+
+Building CNVnator requires at least the following:
+    GNU make
+    C compiler (e.g. gcc or clang)
+    CERN ROOT (https://root.cern.ch)
+    samtools with HTSlib
+
+You must install ROOT package and set up $ROOTSYS variable (see ROOT documentation 
+https://root.cern.ch/root/html534/guides/users-guide/GettingStarted.html).
+
+Download last samtools release https://github.com/samtools/samtools/releases and
+compile samtools and HTSlib.
+
+Building
+========
+
+Download last CNVnator release v0.4.1:
+https://github.com/abyzovlab/CNVnator/releases/download/v0.4.1/CNVnator_v0.4.1.zip
+
+Unzip and compile:
+
+    unzip CNVnator_v0.4.1.zip
+    cd CNVnator_v0.4.1
+    cd src
+    ln -s <path_to_samtools_dir> samtools
+    make -OMP=no # without parallel support
+
+or
+
+    make         # with parallel suport
+
+Installing
+==========
+
+Add CNVnator_v0.4.1/src folder to $PATH by adding following line to .bash_profile:
+
+    export PATH=<path_to_CNVnator_v0.4.1>/src:$PATH
+
+or create symbolic links in folder that is already in $PATH (e.g. ~/bin/):
+
+    cd <local_bin_folder>
+    ln -s <path_to_CNVnator_v0.4.1>/src/cnvnator
+    ln -s <path_to_CNVnator_v0.4.1>/src/plotcircular.py
+    ln -s <path_to_CNVnator_v0.4.1>/src/plotrdbaf.py
+    ln -s <path_to_CNVnator_v0.4.1>/src/plotbaf.py
+    ln -s <path_to_CNVnator_v0.4.1>/src/pytools
+
+Running
+=======
+
+* Extract read mapping:
+
+    cnvnator -root file.root -tree file.bam
+
+* Generate histogram:
+
+    cnvnator -root file.root -his 1000 -fasta reference_genome.fasta.gz -chrom $(seq 1 22) X Y
+or
+    cnvnator -root file.root -his 1000 -fasta reference_genome.fasta.gz -chrom $(seq -f "chr%g" 1 22) chrX chrY
+
+* Calculate statistics
+
+    cnvnator -root file.root -stat 1000 
+
+* Partition
+
+    cnvnator -root file.root -partition 1000
+
+* Call CNVs
+
+    cnvnator -root file.root -call 1000
+
+* Import SNP data
+
+    cnvnator -root file.root -vcf file.vcf.gz
+
+* Import mask data
+
+    cnvnator -root file.root -mask mask.fa.gz
+
+For hg19 reference genome mask file is provaded with CNVnator:
+
+    CNVnator_v0.4.1/ExampleData/20141020.strict_mask.whole_genome.fasta.gz
+
+* Generate SNP histograms
+
+    cnvnator -root file.root -baf 10000
+
+* List root file content
+
+    cnvnator -root file.root -ls
+
+* Copy RD and SNP data to new root file
+
+    cnvnator -root file.root -cptrees new_file.root
+
+* Genotype genomic regions - interactive mode
+
+    cnvnator -root file.root -genotype bin_size
+
+* Ploting - interactive mode
+
+    cnvnator -root file.root -view 1000
+
+* Ploting RD and BAF whole genome circular plots using python tool:
+
+    plotcircular.py file.root
+
+all options:
+
+    plotcircular.py [-chrom CHROMOSOMES] [-bs BINSIZE] [-o SAVE_FILE]
+                    [-t TITLE] [-rdbs RDBINSIZE] [-pbs PLOTBINSIZE]
+                    [-nomask] [-useid] root_file
+
+* Ploting RD and BAF signal over genomic region using python tool:
+
+    plotrdbaf.py file.root 15:1000000-5000000
+
+all options:
+
+    plotrdbaf.py [-h] [-bs BINSIZE] [-rdbs RDBINSIZE] [-res RESOLUTION]
+                 [-o SAVE_FILE] [-t TITLE] [-nomask] [-useid]
+                 root_file region
+
+* Ploting BAF signal over genomic region using python tool:
+
+    plotbaf.py file.root 15:1000000-5000000
+
+all options:
+
+    plotbaf.py [-h] [-bs BINSIZE] [-rdbs RDBINSIZE] [-res RESOLUTION]
+               [-o SAVE_FILE] [-t TITLE] [-nomask] [-useid]
+               root_file region

IO.cpp

@@ -31,7 +31,7 @@ const vector<string> IO::Signal = vector<string>({
 // Distribution use same signal name + " dist"
 
 
-IO::IO(string _rootfile) : rootfile(_rootfile),file(rootfile,"Read") {
+IO::IO(string _rootfile) : rootfile(_rootfile),file(rootfile,"Update") {
   if (file.IsZombie()) {
     cerr<<"Can't open file '"<<rootfile<<"'."<<endl;
     return;
@@ -364,12 +364,13 @@ TH2* IO::newSignalTH2(string title,string chr, int bin, string signal, unsigned
 }
 
 bool IO::writeHistogramsToBinDir(int bin,TH1 *his1,TH1 *his2,TH1 *his3,TH1 *his4,TH1 *his5,TH1 *his6) {
-  TFile file(rootfile,"Update");
-  if(file.IsZombie()) {
-    cerr << "Can't open file '" << rootfile << "'." <<endl;
-    return false;
-  }
+  //TFile file(rootfile,"Update");
+  //if(file.IsZombie()) {
+  //  cerr << "Can't open file '" << rootfile << "'." <<endl;
+  //  return false;
+  //}
   if(bin>0) {
+    file.cd();
     TString dir_name=getDirName(bin);
     TDirectory *dir = (TDirectory*)file.Get(dir_name);
     if (!dir) {
@@ -389,7 +390,7 @@ bool IO::writeHistogramsToBinDir(int bin,TH1 *his1,TH1 *his2,TH1 *his3,TH1 *his4
   if (his4) his4->Write(his4->GetName(),TObject::kOverwrite);
   if (his5) his5->Write(his5->GetName(),TObject::kOverwrite);
   if (his6) his6->Write(his6->GetName(),TObject::kOverwrite);
-  file.Close();
+  //file.Close();
   return true;
 }
 

IO.hh

@@ -29,7 +29,6 @@ class IO {
   const static map<string,unsigned int> Flag;
 
   TString rootfile;
-  TFile file;
   vector<string> rd_tree;
   vector<string> snp_tree;
   vector<string> bin_dir;
@@ -39,7 +38,8 @@ class IO {
 public:
   IO(string _rootfile);
   ~IO();
-
+
+  TFile file;
   // NAMES
   TString treeName(string chr,string signal);
   TString signalName(string chr, int bin, string signal, unsigned int flags);

Makefile

@@ -1,4 +1,4 @@
-VERSION	       = v0.4
+VERSION	       = v0.4.1
 override LIBS += -lz -lbz2 -lcurl -llzma -lreadline -lcrypto
 
 ifneq ($(wildcard $(ROOTSYS)/lib/root),)
@@ -38,19 +38,18 @@ ifneq ($(YEPPPINCLUDEDIR),)
 endif
 
 CXX    = g++ -O3 -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++11 -DCNVNATOR_VERSION=\"$(VERSION)\" $(OMPFLAGS)
-#CXX    = g++ -O3 -DCNVNATOR_VERSION=\"$(VERSION)\" $(OMPFLAGS)
 
 OBJDIR = obj
-OBJS   = $(OBJDIR)/cnvnator.o  \
-	 $(OBJDIR)/EXOnator.o  \
-	 $(OBJDIR)/IO.o  \
-	 $(OBJDIR)/Visualizer.o  \
-	 $(OBJDIR)/HisMaker.o  \
+OBJS   = $(OBJDIR)/cnvnator.o \
+	 $(OBJDIR)/EXOnator.o \
+	 $(OBJDIR)/IO.o \
+	 $(OBJDIR)/Visualizer.o \
+	 $(OBJDIR)/HisMaker.o \
 	 $(OBJDIR)/AliParser.o \
 	 $(OBJDIR)/FastaParser.o \
 	 $(OBJDIR)/VcfParser.o \
 	 $(OBJDIR)/Genotyper.o \
-	 $(OBJDIR)/Interval.o  \
+	 $(OBJDIR)/Interval.o \
 	 $(OBJDIR)/Genome.o
 
 DISTRIBUTION = $(PWD)/CNVnator_$(VERSION).zip
@@ -74,17 +73,19 @@ clean:
 distribution: clean all
 	@echo Creating directory ...
 	@rm -rf $(MAINDIR)
-	@rm -f  $(DISTRIBUTION)
-	@mkdir  $(MAINDIR)
-	@mkdir  $(SRCDIR)
+	@rm -f $(DISTRIBUTION)
+	@mkdir $(MAINDIR)
+	@mkdir $(SRCDIR)
 	@echo Copying files ...
-	@cp *.hh *.cpp  $(SRCDIR)
-	@cp Makefile    $(SRCDIR)
-	@cp README.md       $(MAINDIR)
+	@cp *.hh *.cpp $(SRCDIR)
+	@cp Makefile $(SRCDIR)
+	@cp README.md $(MAINDIR)
 	@cp ReleaseNotes.md $(MAINDIR)
-	@cp CITATION        $(MAINDIR)
-	@cp license.rtf     $(MAINDIR)
-	@cp cnvnator2VCF.pl plotbaf.py $(MAINDIR)
+	@cp INSTALL $(MAINDIR)
+	@cp CITATION $(MAINDIR)
+	@cp license.rtf $(MAINDIR)
+	@cp -r ExampleData $(MAINDIR)
+	@cp -r pytools cnvnator2VCF.pl plotbaf.py plotrdbaf.py plotcircular.py $(SRCDIR)
 	@echo Zipping ...
 	@ln -s $(MAINDIR)
 	@zip -qr $(DISTRIBUTION) $(CNVDIR)

Visualizer.cpp

@@ -166,7 +166,6 @@ char** Visualizer::completer(const char* text, int start, int end) {
 }
 
 char* Visualizer::completion_generator(const char* text, int state) {
-//  std::vector<std::string> vocabulary{"set RD", "set RD raw", "set RD partition", "set RD call", "set SNP", "set SNP likelihood", "set SNP likelihood call", "set SNP count", "show"};
   static std::vector<std::string> matches;
   static size_t match_index = 0;
   if (state == 0) {
@@ -190,7 +189,6 @@ char* Visualizer::completion_generator(const char* text, int state) {
 
 void Visualizer::prompt() {
   TTimer  *timer = new TTimer("gSystem->ProcessEvents();",50,kFALSE);
-//  TString input = "";
   string input="";
   char* buf;
   rl_attempted_completion_function = completer;
@@ -205,17 +203,12 @@ void Visualizer::prompt() {
     input=string(buf);
     if(buf!=nullptr) add_history(buf);
     free(buf);
-//    input = Getline(">");
-//    getline(cin,input);
-//    input.ReplaceAll("\n","\0");
-//    input = input.Remove(TString::kBoth,' ');
     timer->TurnOff();
   }
   delete timer;
 }
 
 void Visualizer::gview() {
-//  int n=io[0]->nChromWithTree();
   int n=chroms.size();
   TCanvas *canvas = new TCanvas("canv","canv",1000,800);
   canvas->SetFillColor(kWhite);
@@ -234,7 +227,6 @@ void Visualizer::generateView(string chrom,int start,int end,TVirtualPad *canvas
   int win = 1*(end - start + 1);
   if (win < 10000) win = 10000;
 
-//  if (canvas) delete canvas;
   TStyle *st = new TStyle("st","st");
   st->SetOptStat(false);  // No box with statistics
   st->SetOptTitle(false); // No box with title
@@ -263,14 +255,6 @@ void Visualizer::generateView(string chrom,int start,int end,TVirtualPad *canvas
     pad->SetFrameLineColor(kWhite);
     pad->SetFrameBorderMode(0);
     drawHistograms(chrom,start,end,win,"RD",pad,raw,his,hisc,hisp,hism);
-//  if (hisc) {
-//    double mean,sigma;
-//    TH1 *rd_his = getHistogram(getDistrName("chr1",bin_size,
-//                                            useATcorr,useGCcorr));
-//    getMeanSigma(rd_his,mean,sigma);
-//    hisc->GetYaxis()->SetRangeUser(0,2*mean);
-//    hisc->SetLineWidth(1);
-//  }
   }
 
   if(panel(2)) {

Visualizer.hh

@@ -4,6 +4,8 @@
 #include "IO.hh"
 #include <map>
 #include <vector>
+#include <string>
+
 
 class Visualizer
 {

cnvnator.cpp

@@ -1,6 +1,7 @@
 // C/C++ includes
 #include <iostream>
 #include <fstream>
+#include <string>
 
 #ifdef USE_YEPPP
 #include <yepMath.h>

plotrdbaf.py

@@ -11,7 +11,7 @@
 parser.add_argument("-bs", "--binsize", type=int,
                     help="size of bins", default=100000)
 parser.add_argument("-rdbs", "--rdbinsize", type=int,
-                    help="size of bins for RD signal", default=10000)
+                    help="size of bins for RD signal", default=100000)
 parser.add_argument("-res", "--resolution", type=int,
                     help="size of bins", default=100)
 parser.add_argument("-o", "--save_file",

pytools/io.py

@@ -123,8 +123,11 @@ def get_signal(self, chr, bin_size, signal, flags=FLAG_USEMASK | FLAG_GC_CORR):
         return [his.GetBinCenter(i) for i in range(n)], [his.GetBinContent(i) for i in range(n)]
 
     def get_signal_2d(self, chr, bin_size, signal, flags=FLAG_USEMASK | FLAG_GC_CORR):
-        """ToDo - Returns 2D histogram: (Xbin_centers, Values) """
-        return True
+        """Returns 2D histogram: (Xbin_centers, Values) """
+        his = self.file.Get("bin_" + str(bin_size)).Get(self.signal_name(chr, bin_size, signal, flags))  
+        nx=his.GetXaxis().GetNbins()
+        ny=his.GetYaxis().GetNbins()
+        return [his.GetXaxis().GetBinCenter(i) for i in range(nx)],[his.GetYaxis().GetBinCenter(i) for i in range(ny)],[[his.GetBinContent(i,j) for j in range(ny)] for i in range(nx)]
 
 
 if __name__ == '__main__':