BioSequence.h

Go to the documentation of this file.

/****************************************************************************
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
The BioBhasha : A Programming Language for Biologist
Version 1.0 (19th December 2001)
Dr. Prasad, B.V.L.S.
Contact: prasadbvls@helixgenomics.com
Version 1.1 (23th January 2003)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
*****************************************************************************/
#ifndef BIOSEQUENCE_H
#define BIOSEQUENCE_H


#include <string>
#include <fstream>
#include <vector>
#include <ctime>
#include "BioUtilities.h"
#include "BioPostScript.h"

using namespace std;

class BioSequence 
{
         friend bool operator >(const BioSequence& , const BioSequence&);
         friend bool operator <(const BioSequence& , const BioSequence&);
         friend bool operator ==(const BioSequence& , const BioSequence&);


protected:

        string seqname_;
        string sequence_;

public:
        BioSequence();
        BioSequence(const string& filename);

        BioSequence(const string& seqnam, const string& seq);

        void setSequence(const string& seqnam, const string& seq);
        void setSequence(const string& seq);
        void setSequenceName(const string& seqnam);
        bool checkSequence();

        int readFasta(ifstream &);
        string getSequenceName() const; 
        string getSequence() const ;
        long getSequenceLength() const ;

        long getNumberOfOccurrences(const char& ch);
        long getNumberOfOccurrences(const string& pattern);
        
        bool findPattern(const string& pat);
        bool find(const string& words);


        string getSequenceSegment(const long& a, const long& b);

        void showPositionsOfPattern(const string& pat, ostream& os=cout);
        void showPositionsOfPattern(const string& pat, BioPostScript& ps,
                  const int& xpos = 100, const int& ypos = 650, const int& w =450 ,const int& h = 20);
        vector<long> getPositionsOfPattern(const string& pat);

        string getMutatedSequence(const long& , const char& );
        string getMutatedSequence(const long&, const string&);
        string getMutatedSequence(const char&,const char&);

        string getMutatedSequence(const string&,const string&);
        string getMutatedSequence(const long&,const long&,const string&);


  
        string getInsertedSequence(const long& , const string&);

        string getDeletedSequence(const long&,const long&);

        void showInFasta(ostream & = cout);

        // Standard Site-Directed Mutation NOTATION
        // P20G, G160A etc...
// string& s1 and NOT const string& s1 because s1 is getting modified inside the function.
// if we declare as const string& , s1 cannot further be modified.

        string getMutatedSequence(string& s1);
        // Ex: string getMutatedSequence("P20G");
        // Ex: string getMutatedSequence("A120T-C34G");
        // A-Single Letter Representation
        // 120-The Site Position where MUtation has to take place
        // T-Single Letter Representation - To be Substituted With.

        long getNumberOfSSRs(const string& pat, const int& repeats);
        vector < long > getPositionsOfSSRs(const string& pat, const int& repeats);

        void showSSRs(const string& pat,const  int& repeats, ostream& os = cout);       
        void showSSRs(const string& pat, const int& repeats, BioPostScript& ps,
                const int& xpos = 100, const int& ypos = 650,const int& w =450 ,const int& h = 20);
        string getShuffledSequence();
        void showShuffledSequence(ostream& os = cout);
};


#endif