#include <boost/serialization/vector.hpp>
#include <boost/shared_ptr.hpp>
+#include <boost/enable_shared_from_this.hpp>
#include <iostream>
-#include "alg/alphabet.hpp"
+#include "alphabet.hpp"
+#include "seq.hpp"
+#include "seq_span.hpp"
// Sequence data class
};
BOOST_CLASS_EXPORT(motif);
-//! the only purpose of this class is that the shared_ptr template
-//! functions need the serialization support to be in-class.
-class seq_string : public std::string
-{
-public:
- typedef std::string::iterator iterator;
- typedef std::string::reverse_iterator reverse_iterator;
- typedef std::string::const_iterator const_iterator;
- typedef std::string::const_reverse_iterator const_reverse_iterator;
-private:
- friend class boost::serialization::access;
- template<class Archive>
- void serialize(Archive& ar, const unsigned int /*version*/) {
- //ar & BOOST_SERIALIZATION_BASE_OBJECT_NVP(std::string);
- ar & boost::serialization::make_nvp("bases",
- boost::serialization::base_object<std::string>(*this)
- );
- }
-};
//! sequence track for mussa.
class Sequence
{
public:
- typedef std::string::value_type value_type;
- typedef std::string::difference_type difference_type;
- typedef std::string::iterator iterator;
- typedef std::string::reverse_iterator reverse_iterator;
- typedef std::string::const_iterator const_iterator;
- typedef std::string::const_reverse_iterator const_reverse_iterator;
- typedef std::string::reference reference;
- typedef std::string::const_reference const_reference;
- typedef std::string::size_type size_type;
- static const size_type npos = std::string::npos;
+ typedef SeqString::value_type value_type;
+ typedef SeqString::difference_type difference_type;
+ typedef SeqString::iterator iterator;
+ typedef SeqString::reverse_iterator reverse_iterator;
+ typedef SeqString::const_iterator const_iterator;
+ typedef SeqString::const_reverse_iterator const_reverse_iterator;
+ typedef SeqString::reference reference;
+ typedef SeqString::const_reference const_reference;
+ typedef SeqString::size_type size_type;
+ static const size_type npos = SeqString::npos;
enum strand_type { UnknownStrand, PlusStrand, MinusStrand, BothStrand };
enum alphabet_ref { reduced_dna_alphabet, reduced_rna_alphabet, reduced_nucleic_alphabet,
nucleic_alphabet, protein_alphabet };
Sequence(const char* seq, alphabet_ref a = reduced_nucleic_alphabet);
Sequence(const std::string& seq, alphabet_ref a = reduced_nucleic_alphabet);
Sequence(const Sequence& seq);
+ Sequence(const Sequence *);
+ Sequence(const SeqSpanRef&, alphabet_ref a = reduced_nucleic_alphabet);
~Sequence();
//! assignment to constant sequences
Sequence &operator=(const Sequence&);
strand_type strand=UnknownStrand);
//! retrive element at specific position
- const_reference at(size_type n) const;
+ const_reference at(size_type i) const { return seq->at(i); }
//! clear the sequence and its annotations
void clear();
- //! return c pointer to the sequence data
- const char *c_str() const;
+ //! return a non-null terminated c pointer to the sequence data
+ const char *data() const { return seq->data(); }
//! forward iterator
- const_iterator begin() const;
+ const_iterator begin() const { return seq->begin(); }
//! last iterator
- const_iterator end() const;
+ const_iterator end() const { return seq->end(); }
//! is our sequence empty?
- bool empty() const;
+ bool empty() const { return (seq) ? seq->empty() : true ; }
+ //! find first
+ size_type find_first_not_of(const std::string& q, size_type index=0) { return seq->find_first_not_of(q, index); }
//! how many base pairs are there in our sequence
- size_type size() const;
+ size_type size() const { return (seq) ? seq->size() : 0; }
//! alias for size (used by string)
- size_type length() const;
+ size_type length() const { return size(); }
+ //! reverse iterator
+ const_reverse_iterator rbegin() const { return seq->rbegin(); }
+ //! reverse end iterator
+ const_reverse_iterator rend() const { return seq->rend(); }
+ //! is our sequence empty?
//! start position relative to "base" sequence
- size_type start() const;
+ size_type start() const { return seq->parentStart(); }
//! one past the last position relative to "base" sequence
- size_type stop() const;
+ size_type stop() const { return seq->parentStop(); }
//! return a subsequence, copying over any appropriate annotation
- Sequence subseq(int start=0, int count = std::string::npos);
+ Sequence subseq(size_type start=0, size_type count = npos) const;
//! reverse a character
std::string create_reverse_map() const;
//! return a reverse compliment (this needs to be improved?)
alphabet_ref a,
int seq_num=1,
int start_index=0, int end_index=0);
- void load_fasta(std::iostream& file,
+ void load_fasta(std::istream& file,
int seq_num=1, int start_index=0, int end_index=0);
//! load sequence from stream
//! \throw mussa_load_error
//! \throw sequence_empty_error
//! \throw sequence_empty_file_error
- void load_fasta(std::iostream& file,
+ void load_fasta(std::istream& file,
alphabet_ref a,
int seq_num=1,
int start_index=0, int end_index=0);
//! load sequence annotations
//! \throws mussa_load_error
void load_annot(std::fstream& data_stream, int start_index, int end_index);
- bool parse_annot(std::string data, int start_index=0, int end_index=0);
+ //! parse annotation file
+ /*! \throws annotation_load_error
+ */
+ void parse_annot(std::string data, int start_index=0, int end_index=0);
//! add an annotation to our list of annotations
void add_annotation(const annot& a);
const std::list<annot>& annotations() const;
std::list<Sequence>::iterator end);
void save(boost::filesystem::fstream &save_file);
- void load_museq(boost::filesystem::path load_file_path, int seq_num);
+ void load_museq(boost::filesystem::path load_file_path, int seq_num);
-private:
- //! parent sequence
- Sequence *parent;
- //! hold a shared pointer to our sequence string
- boost::shared_ptr<seq_string> seq;
+protected:
+ SeqSpanRef seq;
//! which alphabet we're using
alphabet_ref alphabet;
- //! start offset into the sequence
- size_type seq_start;
- //! number of basepairs of the shared sequence we represent
- size_type seq_count;
//! strand orientation
strand_type strand;
//! fasta header
//! a seperate list for motifs since we're currently not saving them
std::list<motif> motif_list;
+ //! copy over all our annotation children
+ void copy_children(Sequence &, size_type start, size_type count) const;
+
void motif_scan(const Sequence& a_motif, std::vector<int> * motif_match_starts) const;
std::string rc_motif(std::string a_motif) const;
//! look for a string sequence type and and it to an annotation list
friend class boost::serialization::access;
template<class Archive>
void serialize(Archive& ar, const unsigned int /*version*/) {
- ar & BOOST_SERIALIZATION_NVP(parent);
ar & BOOST_SERIALIZATION_NVP(seq);
ar & BOOST_SERIALIZATION_NVP(alphabet);
- ar & BOOST_SERIALIZATION_NVP(seq_start);
- ar & BOOST_SERIALIZATION_NVP(seq_count);
ar & BOOST_SERIALIZATION_NVP(strand);
ar & BOOST_SERIALIZATION_NVP(header);
ar & BOOST_SERIALIZATION_NVP(species);
}
};
BOOST_CLASS_EXPORT(Sequence);
-
#endif
projects / mussa.git / blobdiff