X-Git-Url: http://woldlab.caltech.edu/gitweb/?a=blobdiff_plain;f=alg%2Fsequence.cpp;h=521496d3497b2b96d4786819438d170f6ac50ce4;hb=f3c0553a22b0e4ddc39ee45f51725352d92e97f1;hp=608ddd3b2059ba808c283e7726b181bfb27e0ce2;hpb=f3e6762bedaf3775619a36f4103c8cf35d2ca887;p=mussa.git

diff --git a/alg/sequence.cpp b/alg/sequence.cpp
index 608ddd3..521496d 100644
--- a/alg/sequence.cpp
+++ b/alg/sequence.cpp
@@ -32,6 +32,7 @@ namespace fs = boost::filesystem;
 namespace spirit = boost::spirit;
 
 #include "alg/sequence.hpp"
+#include "io.hpp"
 #include "mussa_exceptions.hpp"
 
 #include <string>
@@ -40,27 +41,7 @@ namespace spirit = boost::spirit;
 #include <sstream>
 #include <set>
 
-annot::annot() 
- : begin(0),
-   end(0),
-   type(""),
-   name("")
-{
-}
-
-annot::annot(int begin, int end, std::string type, std::string name)
- : begin(begin),
-   end(end),
-   type(type),
-   name(name)
-{
-}
-
-annot::~annot()
-{
-}
-
-bool operator==(const annot& left, const annot& right)
+bool operator==(const motif& left, const motif& right)
 {
   return ((left.begin== right.begin) and
           (left.end == right.end) and
@@ -68,8 +49,20 @@ bool operator==(const annot& left, const annot& right)
           (left.name == right.name));
 }
 
-motif::motif(int begin, std::string motif)
- : annot(begin, begin+motif.size(), "motif", motif),
+motif::motif() 
+ : begin(0),
+   end(0),
+   type("motif"),
+   name(""),
+   sequence("")
+{
+}
+
+motif::motif(int begin_, std::string motif)
+ : begin(begin_),
+   end(begin_+motif.size()),
+   type("motif"),
+   name(motif),
    sequence(motif)
 {
 }
@@ -78,9 +71,10 @@ motif::~motif()
 {
 }
 
-
 Sequence::Sequence(AlphabetRef alphabet)
-  : seq(new SeqSpan("", alphabet, SeqSpan::PlusStrand))
+  : seq(new SeqSpan("", alphabet, SeqSpan::PlusStrand)),
+    annotation_list(new SeqSpanRefList), 
+    motif_list(new MotifList)
 {
 }
 
@@ -90,7 +84,9 @@ Sequence::~Sequence()
 
 Sequence::Sequence(const char *seq, AlphabetRef alphabet_, SeqSpan::strand_type strand_)
   : header(""),
-    species("")
+    species(""),
+    annotation_list(new SeqSpanRefList),
+    motif_list(new MotifList)
 {
   set_filtered_sequence(seq, alphabet_, 0, npos, strand_);
 }
@@ -99,7 +95,9 @@ Sequence::Sequence(const std::string& seq,
                    AlphabetRef alphabet_,
                    SeqSpan::strand_type strand_) 
   : header(""),
-    species("")
+    species(""),
+    annotation_list(new SeqSpanRefList),
+    motif_list(new MotifList)
 {
   set_filtered_sequence(seq, alphabet_, 0, seq.size(), strand_);
 }
@@ -108,7 +106,7 @@ Sequence::Sequence(const Sequence& o)
   : seq(o.seq),
     header(o.header),
     species(o.species),
-    annots(o.annots),
+    annotation_list(o.annotation_list),
     motif_list(o.motif_list)
 {
 }
@@ -117,7 +115,7 @@ Sequence::Sequence(const Sequence* o)
   : seq(o->seq),
     header(o->header),
     species(o->species),
-    annots(o->annots),
+    annotation_list(o->annotation_list),
     motif_list(o->motif_list)
 {
 }
@@ -126,15 +124,30 @@ Sequence::Sequence(const SequenceRef o)
   : seq(new SeqSpan(o->seq)),
     header(o->header),
     species(o->species),
-    annots(o->annots),
+    annotation_list(new SeqSpanRefList),
     motif_list(o->motif_list)
 {
+  // copy over the annotations in the other sequence ref, 
+  // attaching them to our current sequence ref
+  for(SeqSpanRefList::const_iterator annot_i = o->annotation_list->begin();
+      annot_i != o->annotation_list->end();
+      ++annot_i)
+  {
+    size_type annot_begin= (*annot_i)->start();
+    size_type annot_count = (*annot_i)->size();
+    
+    SeqSpanRef new_annot(seq->subseq(annot_begin, annot_count));
+    new_annot->setAnnotations((*annot_i)->annotations());
+    annotation_list->push_back(new_annot);
+  }  
 }
 
 Sequence::Sequence(const SeqSpanRef& seq_ref)
   : seq(seq_ref),
     header(""),
-    species("")
+    species(""),
+    annotation_list(new SeqSpanRefList),
+    motif_list(new MotifList)
 {
 }
 
@@ -144,28 +157,12 @@ Sequence &Sequence::operator=(const Sequence& s)
     seq = s.seq;
     header = s.header;
     species = s.species;
-    annots = s.annots;
+    annotation_list = s.annotation_list;
     motif_list = s.motif_list;
   }
   return *this;
 }
 
-static void multiplatform_getline(std::istream& in, std::string& line)
-{
-  line.clear();
-  char c;
-  in.get(c);
-  while(in.good() and !(c == '\012' or c == '\015') ) {
-    line.push_back(c);
-    in.get(c);
-  }
-  // if we have cr-lf eat it
-  c = in.peek();
-  if (c=='\012' or c == '\015') {
-    in.get();
-  }
-}
-
 void Sequence::load_fasta(fs::path file_path, int seq_num, int start_index, int end_index)
 {
   load_fasta(file_path, reduced_nucleic_alphabet, seq_num, start_index, end_index);
@@ -327,6 +324,21 @@ Sequence::load_annot(fs::path file_path, int start_index, int end_index)
     throw mussa_load_error("Error loading annotation file " + file_path.string());
   }
 
+  try {  
+    load_annot(data_stream, start_index, end_index);
+  }  catch(annotation_load_error e) {
+    std::ostringstream msg;
+    msg << file_path.native_file_string()
+        << " "
+        << e.what();
+    throw annotation_load_error(msg.str());
+  }
+  data_stream.close();
+}
+
+void
+Sequence::load_annot(std::istream& data_stream, int start_index, int end_index)
+{
   // so i should probably be passing the parse function some iterators
   // but the annotations files are (currently) small, so i think i can 
   // get away with loading the whole file into memory
@@ -336,17 +348,8 @@ Sequence::load_annot(fs::path file_path, int start_index, int end_index)
     data_stream.get(c);
     data.push_back(c);
   }
-  data_stream.close();
 
-  try {  
-    parse_annot(data, start_index, end_index);
-  } catch(annotation_load_error e) {
-    std::ostringstream msg;
-    msg << file_path.native_file_string()
-        << " "
-        << e.what();
-    throw annotation_load_error(msg.str());
-  }
+  parse_annot(data, start_index, end_index);
 }
 
 /* If this works, yikes, this is some brain hurting code.
@@ -367,20 +370,23 @@ Sequence::load_annot(fs::path file_path, int start_index, int end_index)
  */
   
 struct push_back_annot {
-  std::list<annot>& annot_list;
+  Sequence* parent;
+  SeqSpanRefListRef children;
   int& begin;
   int& end;
   std::string& name;
   std::string& type;
   int &parsed;
 
-  push_back_annot(std::list<annot>& annot_list_, 
+  push_back_annot(Sequence* parent_seq,
+                  SeqSpanRefListRef children_list,
                   int& begin_, 
                   int& end_, 
                   std::string& name_, 
                   std::string& type_,
                   int &parsed_) 
-  : annot_list(annot_list_), 
+  : parent(parent_seq),
+    children(children_list),
     begin(begin_),
     end(end_),
     name(name_),
@@ -392,8 +398,7 @@ struct push_back_annot {
   void operator()(std::string::const_iterator, 
                   std::string::const_iterator) const 
   {
-    //std::cout << "adding annot: " << begin << "|" << end << "|" << name << "|" << type << std::endl;
-    annot_list.push_back(annot(begin, end, name, type));
+    children->push_back(parent->make_annotation(name, type, begin, end));
     ++parsed;
   };
 };
@@ -442,8 +447,8 @@ Sequence::parse_annot(std::string data, int start_index, int end_index)
   int end=0;
   std::string name;
   std::string type;
-  std::string seq;
-  std::list<annot> parsed_annots;
+  std::string seqstr;
+  SeqSpanRefListRef parsed_annots(new SeqSpanRefList);
   std::list<Sequence> query_seqs;
   int parsed=0;
 
@@ -485,13 +490,13 @@ Sequence::parse_annot(std::string data, int start_index, int end_index)
                       )
                       // to understand how this group gets set
                       // read the comment above struct push_back_annot
-                     )[push_back_annot(parsed_annots, start, end, type, name, parsed)]
+                     )[push_back_annot(this, parsed_annots, start, end, name, type, parsed)]
                    |
                     ((spirit::ch_p('>')|spirit::str_p("&gt;")) >> 
                        (*(spirit::print_p))[spirit::assign_a(name)] >>
                        spirit::eol_p >> 
-                       (+(spirit::chset<>(Alphabet::nucleic_cstr)))[spirit::assign_a(seq)]
-                     )[push_back_seq(query_seqs, name, seq, parsed)]
+                       (+(spirit::chset<>(Alphabet::nucleic_cstr)))[spirit::assign_a(seqstr)]
+                     )[push_back_seq(query_seqs, name, seqstr, parsed)]
                     ) >>
                     *spirit::space_p
                    )
@@ -502,33 +507,56 @@ Sequence::parse_annot(std::string data, int start_index, int end_index)
     msg << "Error parsing annotation #" << parsed;
     throw annotation_load_error(msg.str());
   }
+  // If everything loaded correctly add the sequences to our annotation list
   // add newly parsed annotations to our sequence
-  std::copy(parsed_annots.begin(), parsed_annots.end(), std::back_inserter(annots));
-  // go seearch for query sequences 
+  std::copy(parsed_annots->begin(), parsed_annots->end(), std::back_inserter(*annotation_list));  
+  // go search for query sequences 
   find_sequences(query_seqs.begin(), query_seqs.end());
 }
 
-void Sequence::add_annotation(const annot& a)
+void Sequence::add_annotation(const SeqSpanRef a)
 {
-  annots.push_back(a);
+  annotation_list->push_back(a);
 }
 
-const std::list<annot>& Sequence::annotations() const
+void Sequence::add_annotation(std::string name, std::string type, size_type start, size_type stop)
 {
-  return annots;
+  add_annotation(make_annotation(name, type, start, stop));
+}
+
+SeqSpanRef 
+Sequence::make_annotation(std::string name, std::string type, size_type start, size_type stop) const
+{
+  // we want things to be in the positive direction
+  if (stop < start) {
+    size_type tmp = start;
+    start = stop;
+    stop = tmp;
+  }
+  size_type count = stop - start;
+  SeqSpanRef new_annot(seq->subseq(start, count,  SeqSpan::UnknownStrand));
+  AnnotationsRef metadata(new Annotations(name));
+  metadata->set("type", type);
+  new_annot->setAnnotations(metadata);
+  return new_annot;
+}
+
+const SeqSpanRefList& Sequence::annotations() const
+{
+  return *annotation_list;
 }
 
 void Sequence::copy_children(Sequence &new_seq, size_type start, size_type count) const
 {
   new_seq.motif_list = motif_list;
-  new_seq.annots.clear();
+  new_seq.annotation_list.reset(new SeqSpanRefList);
 
-  for(std::list<annot>::const_iterator annot_i = annots.begin();
-      annot_i != annots.end();
+  for(SeqSpanRefList::const_iterator annot_i = annotation_list->begin();
+      annot_i != annotation_list->end();
       ++annot_i)
   {
-    size_type annot_begin= annot_i->begin;
-    size_type annot_end = annot_i->end;
+    size_type annot_begin= (*annot_i)->start();
+    size_type annot_end = (*annot_i)->stop();
 
     if (annot_begin < start+count) {
       if (annot_begin >= start) {
@@ -542,9 +570,9 @@ void Sequence::copy_children(Sequence &new_seq, size_type start, size_type count
       } else {
         annot_end = count;
       }
-
-      annot new_annot(annot_begin, annot_end, annot_i->type, annot_i->name);
-      new_seq.annots.push_back(new_annot);
+      SeqSpanRef new_annot(new_seq.seq->subseq(annot_begin, annot_end));
+      new_annot->setAnnotations((*annot_i)->annotations());
+      new_seq.annotation_list->push_back(new_annot);
     }
   }
 }
@@ -558,7 +586,7 @@ Sequence::subseq(size_type start, size_type count, SeqSpan::strand_type strand)
     return new_seq;
   }
 
-  Sequence new_seq = *this;
+  Sequence new_seq(*this);
   new_seq.seq = seq->subseq(start, count, strand);
   if (seq->annotations()) {
     AnnotationsRef a(new Annotations(*(seq->annotations())));
@@ -636,15 +664,18 @@ Sequence::clear()
   seq.reset();
   header.clear();
   species.clear();
-  annots.clear();
-  motif_list.clear();
+  annotation_list.reset(new SeqSpanRefList);
+  motif_list.reset(new MotifList);
 }
 
 void
 Sequence::save(fs::fstream &save_file)
 {
+  std::string type("type");
+  std::string empty_str("");
   //fstream save_file;
-  std::list<annot>::iterator annots_i;
+  SeqSpanRefList::iterator annots_i;
+  AnnotationsRef metadata;
 
   // not sure why, or if i'm doing something wrong, but can't seem to pass
   // file pointers down to this method from the mussa control class
@@ -657,45 +688,149 @@ Sequence::save(fs::fstream &save_file)
 
   save_file << "<Annotations>" << std::endl;
   save_file << species << std::endl;
-  for (annots_i = annots.begin(); annots_i != annots.end(); ++annots_i)
+  for (annots_i = annotation_list->begin(); 
+       annots_i != annotation_list->end(); 
+       ++annots_i)
   {
-    save_file << annots_i->begin << " " << annots_i->end << " " ;
-    save_file << annots_i->name << " " << annots_i->type << std::endl;
+    metadata = (*annots_i)->annotations();
+    save_file << (*annots_i)->parentStart() << " " << (*annots_i)->parentStop() << " " ;
+    save_file << metadata->name() << " " 
+              << metadata->getdefault(type, empty_str) << std::endl;
   }
   save_file << "</Annotations>" << std::endl;
   //save_file.close();
 }
 
-void
-Sequence::load_museq(fs::path load_file_path, int seq_num)
+//void
+//Sequence::load_museq(fs::path load_file_path, int seq_num)
+//{
+//  fs::fstream load_file;
+//  std::string file_data_line;
+//  int seq_counter;
+//  //annot an_annot;
+//  int annot_begin;
+//  int annot_end;
+//  std::string annot_name;
+//  std::string annot_type;
+//  
+//  std::string::size_type space_split_i;
+//  std::string annot_value;
+//
+//  annotation_list.reset(new SeqSpanRefList);
+//  
+//  load_file.open(load_file_path, std::ios::in);
+//
+//  seq_counter = 0;
+//  // search for the seq_num-th sequence 
+//  while ( (!load_file.eof()) && (seq_counter < seq_num) )
+//  {
+//    getline(load_file,file_data_line);
+//    if (file_data_line == "<Sequence>")
+//      seq_counter++;
+//  }
+//  getline(load_file, file_data_line);
+//  // looks like the sequence is written as a single line
+//  set_filtered_sequence(file_data_line, reduced_dna_alphabet, 0, file_data_line.size(), SeqSpan::PlusStrand);
+//  getline(load_file, file_data_line);
+//  getline(load_file, file_data_line);
+//  if (file_data_line == "<Annotations>")
+//  {
+//    getline(load_file, file_data_line);
+//    species = file_data_line;
+//    while ( (!load_file.eof())  && (file_data_line != "</Annotations>") )
+//    {
+//      getline(load_file,file_data_line);
+//      if ((file_data_line != "") && (file_data_line != "</Annotations>"))  
+//      {
+//        // need to get 4 values...almost same code 4 times...
+//        // get annot start index
+//        space_split_i = file_data_line.find(" ");
+//        annot_value = file_data_line.substr(0,space_split_i);
+//        annot_begin = atoi (annot_value.c_str());
+//        file_data_line = file_data_line.substr(space_split_i+1);
+//        // get annot end index
+//        space_split_i = file_data_line.find(" ");
+//        annot_value = file_data_line.substr(0,space_split_i);
+//        annot_end = atoi (annot_value.c_str());
+//
+//        if (space_split_i == std::string::npos)  // no entry for type or name
+//        {
+//          std::cout << "seq, annots - no type or name\n";
+//          annot_name = "";
+//          annot_type = "";
+//        }
+//        else   // else get annot type
+//        {
+//          file_data_line = file_data_line.substr(space_split_i+1);
+//          space_split_i = file_data_line.find(" ");
+//          annot_value = file_data_line.substr(0,space_split_i);
+//          //an_annot.type = annot_value;
+//          annot_type = annot_value;
+//          if (space_split_i == std::string::npos)  // no entry for name
+//          {
+//            std::cout << "seq, annots - no name\n";
+//            annot_name = "";
+//          }
+//          else          // get annot name
+//          {
+//            file_data_line = file_data_line.substr(space_split_i+1);
+//            space_split_i = file_data_line.find(" ");
+//            annot_value = file_data_line.substr(0,space_split_i);
+//            // this seems like its wrong?
+//            annot_type = annot_value;
+//          }
+//        }
+//        add_annotation(annot_name, annot_type, annot_begin, annot_end);
+//      }
+//      //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
+//      //     << "-->" << an_annot.type << "::" << an_annot.name << std::endl;
+//    }
+//  }
+//  load_file.close();
+//}
+
+SequenceRef Sequence::load_museq(boost::filesystem::fstream& load_file)
 {
-  fs::fstream load_file;
+  boost::shared_ptr<Sequence> seq(new Sequence);
   std::string file_data_line;
   int seq_counter;
-  annot an_annot;
+  //annot an_annot;
+  int annot_begin;
+  int annot_end;
+  std::string annot_name;
+  std::string annot_type;
+  
   std::string::size_type space_split_i;
   std::string annot_value;
 
-  annots.clear();
-  load_file.open(load_file_path, std::ios::in);
+  //seq->annotation_list.reset(new SeqSpanRefList);
 
   seq_counter = 0;
-  // search for the seq_num-th sequence 
+  // search for the next sequence
+  int seq_num = 1;
   while ( (!load_file.eof()) && (seq_counter < seq_num) )
   {
     getline(load_file,file_data_line);
     if (file_data_line == "<Sequence>")
       seq_counter++;
   }
+  
+  // Could not find next sequence
+  if (load_file.eof())
+  {
+    seq.reset();
+    return seq;
+  }
+  
   getline(load_file, file_data_line);
   // looks like the sequence is written as a single line
-  set_filtered_sequence(file_data_line, reduced_dna_alphabet, 0, file_data_line.size(), SeqSpan::PlusStrand);
+  seq->set_filtered_sequence(file_data_line, reduced_dna_alphabet, 0, file_data_line.size(), SeqSpan::PlusStrand);
   getline(load_file, file_data_line);
   getline(load_file, file_data_line);
   if (file_data_line == "<Annotations>")
   {
     getline(load_file, file_data_line);
-    species = file_data_line;
+    seq->set_species(file_data_line);
     while ( (!load_file.eof())  && (file_data_line != "</Annotations>") )
     {
       getline(load_file,file_data_line);
@@ -705,45 +840,48 @@ Sequence::load_museq(fs::path load_file_path, int seq_num)
         // get annot start index
         space_split_i = file_data_line.find(" ");
         annot_value = file_data_line.substr(0,space_split_i);
-        an_annot.begin = atoi (annot_value.c_str());
+        annot_begin = atoi (annot_value.c_str());
         file_data_line = file_data_line.substr(space_split_i+1);
         // get annot end index
         space_split_i = file_data_line.find(" ");
         annot_value = file_data_line.substr(0,space_split_i);
-        an_annot.end = atoi (annot_value.c_str());
+        annot_end = atoi (annot_value.c_str());
 
         if (space_split_i == std::string::npos)  // no entry for type or name
         {
           std::cout << "seq, annots - no type or name\n";
-          an_annot.type = "";
-          an_annot.name = "";
+          annot_name = "";
+          annot_type = "";
         }
         else   // else get annot type
         {
           file_data_line = file_data_line.substr(space_split_i+1);
           space_split_i = file_data_line.find(" ");
           annot_value = file_data_line.substr(0,space_split_i);
-          an_annot.type = annot_value;
+          //an_annot.type = annot_value;
+          annot_type = annot_value;
           if (space_split_i == std::string::npos)  // no entry for name
           {
             std::cout << "seq, annots - no name\n";
-            an_annot.name = "";
+            annot_name = "";
           }
           else          // get annot name
           {
             file_data_line = file_data_line.substr(space_split_i+1);
             space_split_i = file_data_line.find(" ");
             annot_value = file_data_line.substr(0,space_split_i);
-            an_annot.type = annot_value;
+            // this seems like its wrong?
+            annot_type = annot_value;
           }
         }
-        annots.push_back(an_annot);  // don't forget to actually add the annot
+        seq->add_annotation(annot_name, annot_type, annot_begin, annot_end);
       }
       //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
       //     << "-->" << an_annot.type << "::" << an_annot.name << std::endl;
     }
   }
-  load_file.close();
+  //load_file.close();
+  return seq;
 }
 
 
@@ -755,18 +893,21 @@ void Sequence::add_motif(const Sequence& a_motif)
       motif_start_i != motif_starts.end();
       ++motif_start_i)
   {
-    motif_list.push_back(motif(*motif_start_i, a_motif.get_sequence()));
+    motif_list->push_back(motif(*motif_start_i, a_motif.get_sequence()));
   }
 }
 
 void Sequence::clear_motifs()
 {
-  motif_list.clear();
+  if (motif_list) 
+    motif_list->clear();
+  else 
+    motif_list.reset(new MotifList);
 }
 
-const std::list<motif>& Sequence::motifs() const
+const Sequence::MotifList& Sequence::motifs() const
 {
-  return motif_list;
+  return *motif_list;
 }
 
 std::vector<int>
@@ -849,7 +990,6 @@ Sequence::motif_scan(const Sequence& a_motif, std::vector<int> * motif_match_sta
     // end Nora stuff, now we see if a match is found this pass
     if (motif_i == motif_len)
     {
-      annot new_motif;
       motif_match_starts->push_back(seq_i - motif_len + 1);
       motif_i = 0;
     }
@@ -864,16 +1004,11 @@ void Sequence::add_string_annotation(std::string a_seq,
 {
   std::vector<int> seq_starts = find_motif(a_seq);
   
-  //std::cout << "searching for " << a_seq << " found " << seq_starts.size() << std::endl;
-
   for(std::vector<int>::iterator seq_start_i = seq_starts.begin();
       seq_start_i != seq_starts.end();
       ++seq_start_i)
   {
-    annots.push_back(annot(*seq_start_i, 
-                           *seq_start_i+a_seq.size(),
-                           "",
-                           name));
+    add_annotation(name, "", *seq_start_i, *seq_start_i+a_seq.size()); 
   }
 }