Changeset 1020

Timestamp:

Apr 3, 2009, 11:53:27 AM (14 years ago)

Author:

Jari Häkkinen

Message:

Addresses #118 and fixes #187.

File:

• plugins/base2/net.sf.basedb.normalizers/trunk/src/c++/bin/qQN.cc (modified) (5 diffs)

plugins/base2/net.sf.basedb.normalizers/trunk/src/c++/bin/qQN.cc

@@ -37 +37 @@
 #include <fstream>
 #include <iostream>
+#include <stdexcept>
 
-using namespace theplu::yat;
+using namespace theplu::yat::normalizer;
+using namespace theplu::yat::utility;
+
+void create_target(std::vector<double>&, const Matrix&);
+void create_target(std::vector<double>&, const Matrix&, const std::string&);
+
 
 int main(int argc, char* argv[])
-{  
-  using namespace normalizer;
-  using namespace utility;
-
+{
   CommandLine cmd;
-  OptionArg<std::string> dir(cmd, "assay-data", "assay annotation file");
+  OptionArg<std::string> assay(cmd, "assay-data", "assay annotation file");
   OptionHelp help(cmd);
   help.synopsis()=(std::string("See ") +
@@ -53 +56 @@
   OptionSwitch version(cmd, "version", "output version and exit");
   std::stringstream copyright;
-  copyright << cmd.program_name() << VERSION << '\n'
+  copyright << PACKAGE_STRING << '\n'
             << "Copyright (C) 2009 Jari Häkkinen\n\n"
             << "This is free software see the source for copying "
@@ -68 +71 @@
     std::cout << e.what() << std::endl;
     return EXIT_FAILURE;
-  }  
+  }
   if (version.present()) {
     std::cout << copyright.str();
@@ -75 +78 @@
 
   Matrix m(std::cin,'\t');
+  std::vector<double> target(m.rows(),0);
+  ( assay.present() ? create_target(target,m,assay.value()) :
+                      create_target(target,m) );
+  qQuantileNormalizer qqn(target.begin(), target.end(), 100);
+  ColumnNormalizer<qQuantileNormalizer> cn(qqn);
   Matrix result(m.rows(),m.columns());
-  qQuantileNormalizer qqn(m.begin_column(0), m.end_column(0),100);
-  ColumnNormalizer<qQuantileNormalizer> cn(qqn);
   cn(m,result);
   std::cout << result << std::endl;
@@ -83 +89 @@
   return EXIT_SUCCESS;
 }
+
+
+void create_target(std::vector<double>& t, const Matrix& m,
+                   const std::string& assay)
+{
+  std::ifstream is(assay.c_str());
+  std::string line;
+  size_t column=0;
+  size_t yes=0;
+  for (size_t row=0; row<m.rows(); ++row)
+    t[row]=0;
+  while (getline(is, line)) {
+    size_t found=line.find("yes");
+    if (found!=std::string::npos) {
+      for (size_t row=0; row<m.rows(); ++row)
+        t[row]+=m(row,column);
+      ++yes;
+    }
+    ++column;
+    if (column>m.columns())
+      throw std::runtime_error("Too many annotation columns wrt data matrix");
+  }
+  if (!yes)
+    throw std::runtime_error("No columns marked as reference");
+  for (size_t row=0; row<m.rows(); ++row)
+    t[row]/=yes;
+}
+
+
+void create_target(std::vector<double>& t, const Matrix& m)
+{
+  for (size_t row=0; row<m.rows(); ++row) {
+    t[row]=0;
+    for (size_t column=0; column<m.columns(); ++column)
+      t[row]+=m(row,column);
+    t[row]/=m.columns();
+  }
+}