From: Diane Trout <diane@caltech.edu>
Date: Wed, 11 Oct 2006 21:37:48 +0000 (+0000)
Subject: add Mussa::save_motif()
X-Git-Url: http://woldlab.caltech.edu/gitweb/?p=mussa.git;a=commitdiff_plain;h=c760a5d3839b241f2f94437731050622b197bf03

add Mussa::save_motif()
necessary for ticket:133
This adds the low level motif saving code to Mussa. To make things
work a bit more reliably, the parse code can understand the alpha
channel now (so I can save hidden motifs), and can parse a
motif name that is quoted like "foo asdf". I added that as I expect
users will put spaces in their motif names.

Although things'll probably break if someone puts a " in a motif name.
see ticket:149
---

diff --git a/alg/color.cpp b/alg/color.cpp
index cbcc59e..64eb782 100644
--- a/alg/color.cpp
+++ b/alg/color.cpp
@@ -90,9 +90,6 @@ bool operator==(const Color &x, const Color &y)
 
 std::ostream &operator<<(std::ostream &out, const Color &c)
 {
-  out << "Color(" << c.r() << ", " 
-                    << c.g() << ", " 
-                    << c.b() << ", " 
-                    << c.a() << ")";
+  out << c.r() << " " << c.g() << " " << c.b() << " " << c.a();
   return out;
 }
diff --git a/alg/mussa.cpp b/alg/mussa.cpp
index 634cb6d..18539b8 100644
--- a/alg/mussa.cpp
+++ b/alg/mussa.cpp
@@ -802,8 +802,9 @@ struct push_back_motif {
     name.clear();
     // be nice if glsequence was a subclass of sequence so we could
     // just attach colors directly to the motif.
-    Color c(red, green, blue);
+    Color c(red, green, blue, alpha);
     color_mapper->appendInstanceColor("motif", seq.c_str(), c);
+    alpha = 1.0;
     motifs.insert(seq);
     ++parsed;
   };
@@ -839,19 +840,29 @@ void Mussa::load_motifs(std::istream &in)
   }
   // parse our string
   bool ok = spirit::parse(data.begin(), data.end(),
-     *( 
+     *(
        ( 
         (
          (+spirit::chset<>(alphabet))[spirit::assign_a(seq)] >> 
          +spirit::space_p
         ) >>
         !(
-          (spirit::alpha_p >> *spirit::graph_p)[spirit::assign_a(name)]
-          >> +spirit::space_p
+          (
+            // names can either be letter followed by non-space characters
+            (spirit::alpha_p >> *spirit::graph_p)[spirit::assign_a(name)]
+            |
+            // or a quoted string
+            (
+             spirit::ch_p('"') >> 
+               (+(~spirit::ch_p('"')))[spirit::assign_a(name)] >>
+             spirit::ch_p('"')
+            )
+          ) >> +spirit::space_p
         ) >>
         spirit::real_p[spirit::assign_a(red)] >> +spirit::space_p >>
         spirit::real_p[spirit::assign_a(green)] >> +spirit::space_p >>
-        spirit::real_p[spirit::assign_a(blue)] >> +spirit::space_p
+        spirit::real_p[spirit::assign_a(blue)] >> +spirit::space_p >>
+        !(spirit::real_p[spirit::assign_a(alpha)] >> +spirit::space_p)
        )[push_back_motif(motif_sequences, color_mapper, seq, name, red, green, blue, alpha, parsed)]
      )).full;
   if (not ok) {
@@ -864,6 +875,28 @@ void Mussa::load_motifs(std::istream &in)
   update_sequences_motifs();
 }
 
+void Mussa::save_motifs(fs::path filename)
+{
+  fs::ofstream out_stream;
+  out_stream.open(filename, ofstream::out);
+  save_motifs(out_stream);
+}
+
+void Mussa::save_motifs(std::ostream& out)
+{
+  for(motif_set::iterator motif_i = motif_sequences.begin();
+      motif_i != motif_sequences.end();
+      ++motif_i)
+  {
+    out << motif_i->get_sequence() << " ";
+    if (motif_i->get_name().size() > 0) {
+      out << "\"" << motif_i->get_name() << "\" ";
+    }
+    out << color_mapper->lookup("motif", motif_i->get_sequence());
+    out << std::endl;
+  }
+}
+
 void Mussa::update_sequences_motifs()
 {
   // once we've loaded all the motifs from the file, 
diff --git a/alg/mussa.hpp b/alg/mussa.hpp
index 64f4649..27cd239 100644
--- a/alg/mussa.hpp
+++ b/alg/mussa.hpp
@@ -181,6 +181,10 @@ public:
     /*! \sa Mussa::load_motifs(boost::filesystem::path)
      */
     void load_motifs(std::istream &);
+    //! save motif list to the specified filename
+    void save_motifs(boost::filesystem::path filename);
+    //! save motif list to an ostream
+    void save_motifs(std::ostream &);
     //! return our motifs;
     const motif_set& motifs() const;
 
diff --git a/alg/test/test_mussa.cpp b/alg/test/test_mussa.cpp
index 74f3c20..cfa8735 100644
--- a/alg/test/test_mussa.cpp
+++ b/alg/test/test_mussa.cpp
@@ -180,7 +180,7 @@ BOOST_AUTO_TEST_CASE( mussa_load_analysis )
 BOOST_AUTO_TEST_CASE( mussa_load_motif )
 {
   string data = "AAGG 1.0 1.0 0.0\n"
-                "GGTT 0.0 0.1 1.0\n";
+                "GGTT 0.0 0.1 1.0 1.0\n";
 
   istringstream test_istream(data);
 
@@ -243,6 +243,74 @@ BOOST_AUTO_TEST_CASE( mussa_weirdly_spaced_named_motif )
   BOOST_REQUIRE_EQUAL(motifs.size(), 1);
   BOOST_CHECK_EQUAL(motifs.begin()->get_name(), "cat_meow123");
 }
+
+BOOST_AUTO_TEST_CASE( mussa_name_quoted_motif )
+{
+  string data = "CCAATT       \"cat meow 123\"     0.1    0.2 0.3\n";
+  istringstream test_istream(data);
+
+  Mussa m1;
+  m1.append_sequence("AAAAGGGGTTTT");
+  m1.append_sequence("GGGCCCCTTCCAATT");
+  m1.load_motifs(test_istream);
+
+  std::set<Sequence> motifs = m1.motifs();
+  BOOST_REQUIRE_EQUAL(motifs.size(), 1);
+  BOOST_CHECK_EQUAL(motifs.begin()->get_name(), "cat meow 123");
+}
+
+BOOST_AUTO_TEST_CASE( mussa_name_embedded_quote_motif )
+{
+  // pretty obviously this shouldn't work as " are our delimiter
+  // and i'm too lazy to add support for \ in the parser
+  string data = "ATA 0.5 0.5 0.5\n"
+                "CCAATT       \"cat \"meow 123\"     0.1    0.2 0.3\n";
+  istringstream test_istream(data);
+
+  Mussa m1;
+  m1.append_sequence("AAAAGGGGTTTT");
+  m1.append_sequence("GGGCCCCTTCCAATT");
+  BOOST_CHECK_THROW( m1.load_motifs(test_istream), motif_load_error);
+
+  std::set<Sequence> motifs = m1.motifs();
+  BOOST_REQUIRE_EQUAL(motifs.size(), 0);
+}
+
+BOOST_AUTO_TEST_CASE( mussa_save_motif )
+{
+  string data = "ATA 1 1 1 1\n"
+                "CAT \"my name\" 1 0 0.5 0.5\n";
+  istringstream data_istream(data);
+
+  Mussa m1;
+  m1.append_sequence("AAAAGGGGTTTT");
+  m1.append_sequence("GGGCCCCTTCCAATT");
+  m1.load_motifs(data_istream);
+  
+  string save;
+  ostringstream save_ostream(save);
+  m1.save_motifs(save_ostream);
+
+  istringstream reloaded_istream(save_ostream.str());
+  Mussa m2;
+  m2.append_sequence("AAAAGGGGTTTT");
+  m2.append_sequence("GGGCCCCTTCCAATT");
+  m2.load_motifs(reloaded_istream);
+  
+  BOOST_REQUIRE_EQUAL(m1.motifs().size(), m2.motifs().size());
+  Mussa::motif_set::const_iterator m1motif = m1.motifs().begin();
+  Mussa::motif_set::const_iterator m2motif = m2.motifs().begin();
+  for (;
+       m1motif != m1.motifs().end() and m2motif != m2.motifs().end();
+       ++m1motif, ++m2motif) 
+  {
+    BOOST_CHECK_EQUAL(m1motif->get_sequence(), m2motif->get_sequence());
+    BOOST_CHECK_EQUAL(m1motif->get_name(), m2motif->get_name());
+    BOOST_CHECK_EQUAL(m1.colorMapper()->lookup("motif", m1motif->get_sequence()),
+                      m2.colorMapper()->lookup("motif", m2motif->get_sequence()));
+  }  
+}
+
 BOOST_AUTO_TEST_CASE( mussa_add_motif )
 {
   vector<Sequence> motifs;