Changeset 1164 for plugins/base2/net.sf.basedb.normalizers/trunk/src/c++/bin/qQN.cc

Timestamp:

Sep 23, 2009, 11:08:19 AM (14 years ago)

Author:

Jari Häkkinen

Message:

Fixes #247. A work around for the erroneous export from exportPlainMatrix.

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

@@ -59 +59 @@
 
 
+void create_matrix_until_183_fixed(MatrixWeighted**, const MatrixWeighted&,
+                                   const std::string&);
 void create_target(std::vector<double>&, const MatrixWeighted&, bool use_median);
 void create_target(std::vector<double>&, const MatrixWeighted&,
@@ -111 +113 @@
     std::cin.rdbuf(infile->rdbuf());
   }
-  MatrixWeighted m(std::cin,'\t');
+
+  // The BASE core API deprecated exportPlainMatrix method of the
+  // deprecated BioAssaySetExporter does not behave well when
+  // exporting bioassaysets where assays have been filtered out in
+  // some filter step. Variables and functions with 183 in their name
+  // should be removed when http://baseplugins.thep.lu.se/ticket/183
+  // is fixed.
+
+  // m is a pointer below, should be changed to a normal variable when
+  // http://baseplugins.thep.lu.se/ticket/183 is fixed.
+
+  // Change to m when http://baseplugins.thep.lu.se/ticket/183 fixed.
+  MatrixWeighted m_tmp_until_183_fixed(std::cin,'\t');
   if (indata.present()) {
     std::cin.rdbuf(cin_buffer); // restore old input buffer
@@ -117 +131 @@
     delete infile;
   }
-
-  std::transform(m.begin(), m.end(), m.begin(), CleanUpMatrix());
+  // Remove until HERE below when
+  // http://baseplugins.thep.lu.se/ticket/183 fixed.
+  MatrixWeighted* m=NULL;
+  if (assay.present())
+    create_matrix_until_183_fixed(&m, m_tmp_until_183_fixed, assay.value());
+  else
+    m = &m_tmp_until_183_fixed;
+  // HERE remove until here.
+
+  std::transform(m->begin(), m->end(), m->begin(), CleanUpMatrix());
   std::vector<double> target;
   ( assay.present() ?
-    create_target(target, m, assay.value(), target_median.value()) :
-    create_target(target, m, target_median.value()) );
+    create_target(target, *m, assay.value(), target_median.value()) :
+    create_target(target, *m, target_median.value()) );
   std::transform(target.begin(), target.end(),
                  target.begin(), theplu::yat::utility::Log<double>());
-  std::transform(data_iterator(m.begin()), data_iterator(m.end()),
-                 data_iterator(m.begin()), theplu::yat::utility::Log<double>());
+  std::transform(data_iterator(m->begin()), data_iterator(m->end()),
+                 data_iterator(m->begin()), theplu::yat::utility::Log<double>());
   // q = min(100,target_size/10) but no smaller than 10
   unsigned int q=target.size()/10;
@@ -132 +154 @@
   qQuantileNormalizer qqn(target.begin(), target.end(), q);
   ColumnNormalizer<qQuantileNormalizer> cn(qqn);
-  MatrixWeighted result(m.rows(),m.columns());
-  cn(m,result);
+  MatrixWeighted result(m->rows(),m->columns());
+  cn(*m,result);
   std::transform(data_iterator(result.begin()), data_iterator(result.end()),
                  data_iterator(result.begin()), theplu::yat::utility::Exp<double>());
@@ -151 +173 @@
   }
 
+  // clean up until http://baseplugins.thep.lu.se/ticket/183 fixed.
+  if (assay.present())
+    delete m;
+
   return EXIT_SUCCESS;
+}
+
+
+void create_matrix_until_183_fixed(MatrixWeighted** m,
+                                   const MatrixWeighted& m_tmp_183,
+                                   const std::string& assay)
+{
+  std::ifstream is(assay.c_str());
+  std::vector<size_t> idx;
+  while (is.good()) {
+    size_t i;
+    is >> i;
+    if (is.good()) idx.push_back(i);
+    std::string line;
+    getline(is, line);
+    if (idx.size()>m_tmp_183.columns())
+      throw std::runtime_error(std::string("183_tmp: Too many annotation ") +
+                               "columns wrt data matrix");
+  }
+
+  if (*m) delete *m;
+  size_t rows=m_tmp_183.rows();
+  size_t cols=idx.size();
+  *m=new MatrixWeighted(rows,cols);
+  for (size_t i=0; i<rows; ++i)
+    for (size_t j=0; j<cols; ++j)
+      (**m)(i,j)=m_tmp_183(i,idx[j]-1);
 }