BioMultipleSequenceAlignment.h

Go to the documentation of this file.

#ifndef BIOMULTIPLESEQUENCEALIGNMENT_H
#define BIOMULTIPLESEQUENCEALIGNMENT_H

#include "BioMultipleFasta.h"
#include "BioOutputStream.h"
#include "BioMultipleGenBank.h"
#include "BioMultipleEmbl.h"
#include "BioMultipleSwissProt.h"
#include "BioMatrix.h"
#include "BioDnaSubstitutionMatrix.h"
#include "BioProteinSubstitutionMatrix.h"

#include <unistd.h>
#include <algorithm>
#include <vector>
#include <stdio.h>
#include<ctype.h>
#include <string.h>
#include <stdlib.h>
#include <getopt.h>
#define SEEK_START 0
#define SEEK_END 2
#define INFTY 0x8000000


struct sequence_info{
        int** s;
        int* sl;
        int** sn;
        int* lsn;
        int** gis;
        int** relpos;
        int** sip;
        int* nsip;
  };

  struct dp_matrix{
        int** tb;
        int** m;
        int* a;
        int* ga;
        int* gb;
        int* true_x;
        int* true_y;
        void* tb_mem;
        void* m_mem;
        int x;
        int y;
  };

class BioMultipleSequenceAlignment
{
  vector<BioFasta> aln_seqs;
  vector<BioFasta> inp_seqs;
  vector<BioFasta> do_complete_alignment(BioMultipleFasta&);
  void* tmalloc(int size);
  struct dp_matrix* dp_matrix_alloc(struct dp_matrix *dp,int x,int y,int p);
  struct dp_matrix* dp_matrix_realloc(struct dp_matrix *dp,int x,int y,int p);
  struct dp_matrix* dp_matrix_init(struct dp_matrix *dp,int x,int y);
  void dp_matrix_free(struct dp_matrix *dp,int p);
  static short int gon250mt[276];
  //struct sequence_info *si2;
  struct sequence_info *si;
  int numseq;// = 0;
  int numprofiles;// = 0;

  int** ten_wu_manber(int* seq,int len,int p[]);

  struct sequence_info* read_sequences(struct sequence_info* si,char* infile);
  double distance_calculation(int** matches[],int len_a,int len_b,int a,int b);
  double distance_calculation2(int** matches[],int len_a,int len_b,int a,int b);

  void fill_hash(struct sequence_info* si,int** matches[],int nowu);
  int* upgma(double **dm,int* tree);
  void add_ptm(int** matches[],int** matrix,int a,int b);
  struct dp_matrix* consistency_check(struct dp_matrix *dp,int len_a,int len_b,int dia);
  int* make_profile(int* prof,int* seq,int len,int** subm);
  int** read_matrix(short *matrix_pointer,int** subm,int gpo);
  int* main_fast_dyn(int* path,struct dp_matrix *dp,int* prof1,int* prof2,int len_a,int len_b);

  int** make_new_profile(int**newp,int** profa,int** profb,int* path);

  vector<BioFasta> print_alignment(struct sequence_info* si);//,char* outfile);
  struct sequence_info* update(struct sequence_info* si,int** profile,int a,int b,int newnode,int* path);

  void set_gap_penalties(int* prof,int len,int nsip);
  void update_gaps(int old_len,int*gis,int new_len,int *newgaps);
  void update_hash(struct sequence_info* si,int** matches[],int a,int b,int enew);
  int gpo; //= 0;
  void add_ptm2(int** matches[],struct dp_matrix *dp,int a,int b);
  struct dp_matrix* consistency_check2(struct dp_matrix *dp,int len_a,int len_b,int dia);
 // char* get_input_into_string(char* string_,string);
  
public:
   //static short int gon250mt;
        BioMultipleSequenceAlignment();
                
        BioMultipleSequenceAlignment(BioMultipleFasta&);
        BioMultipleSequenceAlignment(BioMultipleGenBank&);
        BioMultipleSequenceAlignment(BioMultipleEmbl&);
        BioMultipleSequenceAlignment(BioMultipleSwissProt&);
        void showAlignment(ostream& = cout);
        string getAlignedSequence(int i);
        vector<string> getAlignedSequences();
        string getAlignedSequenceName(int i);
        BioFasta getAlignedSequenceInFasta(int i);
        BioMultipleFasta getAlignedSequencesInMultipleFasta();
        vector<char> getColumn(int i);
        BioMatrix getProfileMatrix(string s);
        int getColumnScore(int col_index, string s);
        vector<int> getColumnScores(string type);
};

#endif