boxchain.h

Go to the documentation of this file.

#ifndef BOXCHAIN_H
#define BOXCHAIN_H

#include <string>
//#include <vector>
#include <iostream>
#include <list>
#include <set>
#include <map>
#include "bioseq.h"
#include "alignseg.h"
#include "matrix.h"
#include "kzblast.h"
#include "dynaln.h"

using namespace std;



//class ltMatchQueryBeginPtr {
class ltAlignsegexBeginYPtr {
   public:
      bool operator()(const Alignsegex *p1, const Alignsegex *p2) const {
         if (p1->getY() < p2->getY()) return true;
         else if (p1->getY() == p2->getY()) {
            if (p1->getScore()<p2->getScore()) return true;
            else return false;
         }
         else return false;
      }
};

class ltAlignsegexEndYPtr {
   public:
      bool operator()(const Alignsegex *p1, const Alignsegex *p2) const {
         if (p1->getEndY() < p2->getEndY()) return true;
         else if (p1->getEndY() == p2->getEndY()) {
            if (p1->getMaxScore()<p2->getMaxScore()) return true;
            else return false;
         }
         else return false;
      }
};

class Boxchain {
   public:
      Boxchain(Alignsegex* pe, const bioseq* xx, const bioseq* yy);
      ~Boxchain();
      friend ostream& operator<<(ostream& ous, const Boxchain& bc);
      void buildAlignment(Dynaln& aln);
      void printAlign(ostream &ous, const int width=70);
      int getXStart() const { return boxes.front()->getX(); }
      int getYStart() const { return boxes.front()->getY(); }
      int getXEnd() const { return boxes.back()->getEndX(); }
      int getYEnd() const { return boxes.back()->getEndY(); }
      int numberHSP() const { return boxes.size(); }
      int getScore() const { return sumscore; }

   private:
      list<Alignsegex*> boxes;
      const bioseq* xseq; // passed on from the outside
      const bioseq* yseq; // from the outside
      string top, buttom, middle;
      int sumscore;
};

class Linkbox {
   public:
      ~Linkbox();
      Linkbox(const list<Alignseg> &asg, const bioseq* qseq, const bioseq* tseq);
      typedef multiset<Alignsegex*, ltAlignsegexEndYPtr>::iterator bestiterator;
      typedef multiset<Alignsegex*, ltAlignsegexEndYPtr>::reverse_iterator bestreverse_iterator;
      typedef multimap<int, Alignsegex*>::iterator mmiterator;

      Boxchain* bestChain(Dynaln& aln);
      int connect(Alignsegex* as1, Alignsegex* as2, Dynaln& aln);

      /* display the final best set content
       * this contains the final result
       */
      //void showBest(ostream &ous) const;
      /* show the best set content directly */
      void show_best(ostream &ous) const;
      //void show_end2match(ostream &ous) const;
      void show_start2match(ostream &ous) const;

      /* find sections on the target (genomic) where all matches
       * are in the same direction as regards to targets
       * Thse section may overlap.
       * Requirement: The tblastn must have been sorted by target.
       * This class does not have enough information to do
       * this job well. The data structure is also not easy
       * for deletion.
       */

      void removeChain(Alignsegex* p);
      void repairEnds(const int window=4, const float frac=0.15);
      static void setMatrix(const Matrix* mt) { matrix=mt; }
      static void setBlastParameter(const BlastParameter* p) { blparam=p; }

   private:
      set<int> xpoints;
      /* only this one holds the data 
       */
      multimap<int, Alignsegex*> start2match;
      multimap<int, Alignsegex*> end2match;
      /* although this set may contain elements starting
       * at the same b.y, their scores must be different
       * Order by queryBegin, then by maxScore
       */
      //multimap<int, Alignsegex*> best;
      multiset<Alignsegex*, ltAlignsegexEndYPtr> best;
      /* bi is the iterator pointing to the 
       * element just inserted.  This element will
       * be used for comparision, any element after
       * this one that has a smaller score will be
       * removed
       */
      //void cleanUpBest(bestiterator bi);
      void cleanUpBest(bestiterator bi);
      const bioseq* xseq; 
      const bioseq* yseq;

      static const BlastParameter* blparam;
      static const Matrix* matrix;
};

#endif