segment.h 5.08 KB
Newer Older
Mikaël Salson's avatar
Mikaël Salson committed
1 2 3 4 5 6 7 8
#ifndef SEGMENT_H
#define SEGMENT_H

#include <string>
#include <iostream>
#include "fasta.h"
#include "dynprog.h"
#include "tools.h"
9
#include "germline.h"
Mikaël Salson's avatar
Mikaël Salson committed
10 11
#include "kmerstore.h"
#include "kmeraffect.h"
Mathieu Giraud's avatar
Mathieu Giraud committed
12 13
#include "affectanalyser.h"
#include "json.h"
Mikaël Salson's avatar
Mikaël Salson committed
14 15

#define EXTEND_D_ZONE 5
Mikaël Salson's avatar
Mikaël Salson committed
16 17 18
#define RATIO_STRAND 5          /* The ratio between the affectations in one
                                   strand and the other, to safely attribute a
                                   segment to a given strand */
Mikaël Salson's avatar
Mikaël Salson committed
19

20
#define DETECT_THRESHOLD 5      /* If the number of both V and J affectations
21 22 23 24
                                   is above this threshold, then the sequence
                                   will be labeled as 'detected', and, if it
                                   not segmented, the remaining germlines will
                                   not be tested */
25

Mathieu Giraud's avatar
Mathieu Giraud committed
26 27
#define JSON_REMEMBER_BEST  4   /* The number of V/D/J predictions to keep  */

Mikaël Salson's avatar
Mikaël Salson committed
28 29 30 31
using namespace std;

enum SEGMENTED { DONT_KNOW, SEG_PLUS, SEG_MINUS, UNSEG_TOO_SHORT, UNSEG_STRAND_NOT_CONSISTENT, 
		 UNSEG_TOO_FEW_ZERO,  UNSEG_TOO_FEW_V, UNSEG_TOO_FEW_J, 
32
		 UNSEG_BAD_DELTA_MIN, UNSEG_BAD_DELTA_MAX, UNSEG_AMBIGUOUS,
Mikaël Salson's avatar
Mikaël Salson committed
33 34 35 36
		 TOTAL_SEG_AND_WINDOW, 
		 TOTAL_SEG_BUT_TOO_SHORT_FOR_THE_WINDOW,
		 STATS_SIZE } ;
const char* const segmented_mesg[] = { "?", "SEG_+", "SEG_-", "UNSEG too short", "UNSEG strand",  
Mikaël Salson's avatar
Mikaël Salson committed
37
				       "UNSEG too few (zero)", "UNSEG too few V", "UNSEG too few J",
38
				       "UNSEG < delta_min", "UNSEG > delta_max", "UNSEG ambiguous",
Mikaël Salson's avatar
Mikaël Salson committed
39 40 41
                                       "= SEG, with window",
                                       "= SEG, but no window",
                                      } ;
Mikaël Salson's avatar
Mikaël Salson committed
42 43 44 45 46

class Segmenter {
protected:
  string label;
  string sequence;
Mathieu Giraud's avatar
Mathieu Giraud committed
47 48
  int Vend, Jstart;
  int Dstart, Dend;
Mikaël Salson's avatar
Mikaël Salson committed
49 50 51 52 53 54 55 56
  bool reversed, segmented, dSegmented;

  string removeChevauchement();
  bool finishSegmentation();
  bool finishSegmentationD();

 public:
  string code;
Mikaël Salson's avatar
Mikaël Salson committed
57
  string code_short;
Mikaël Salson's avatar
Mikaël Salson committed
58
  string code_light;
59 60
  string info;        // .vdj.fa header, fixed fields
  string info_extra;  // .vdj.fa header, other information, at the end of the header
Mikaël Salson's avatar
Mikaël Salson committed
61 62
  int best_V, best_J ;
  int del_V, del_D_left, del_D_right, del_J ;
63
  string seg_V, seg_N, seg_J, system;
64

Mikaël Salson's avatar
Mikaël Salson committed
65 66 67 68
  int best_D;
  string seg_N1, seg_D, seg_N2;
  Cost segment_cost;

Mathieu Giraud's avatar
Mathieu Giraud committed
69
  virtual ~Segmenter();
Mikaël Salson's avatar
Mikaël Salson committed
70

Mathieu Giraud's avatar
Mathieu Giraud committed
71
  /* Queries */
Mikaël Salson's avatar
Mikaël Salson committed
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134



  Sequence getSequence() const ;

  /**
   * @param l: length around the junction
   * @return the string centered on the junction (ie. at position
   *         (getLeft() + getRight())/2).
   *         The string has length l unless the original string 
   *         is not long enough.
   *         The junction is revcomp-ed if the original string comes from reverse
   *         strand.
   */
  string getJunction(int l) const;

  /**
   * @return the left position (on forward strand) of the segmentation.
   */
  int getLeft() const;
  
  /**
   * @return the right position (on forward strand) of the segmentation
   */
  int getRight() const;
  
  /**
   * @return the left position (on forward strand) of the D segmentation.
   */
  int getLeftD() const;
  
  /**
   * @return the right position (on forward strand) of the D segmentation
   */
  int getRightD() const;

  /**
   * @return true iff the string comes from reverse strand
   */
  bool isReverse() const;

  /**
   * @return true iff the sequence has been successfully segmented
   */
  bool isSegmented() const;
  
  /**
   * @return true if a D gene was found in the N region
   */
  bool isDSegmented() const;


  friend ostream &operator<<(ostream &out, const Segmenter &s);
};



ostream &operator<<(ostream &out, const Segmenter &s);



class KmerSegmenter : public Segmenter
{
Mathieu Giraud's avatar
Mathieu Giraud committed
135
 private:
136
  int detected;
Mathieu Giraud's avatar
Mathieu Giraud committed
137
  int because;                  
138
  KmerAffectAnalyser *kaa;
Mikaël Salson's avatar
Mikaël Salson committed
139 140 141 142
 protected:
  string affects;

 public:
143 144
  Germline *segmented_germline;
  
Mikaël Salson's avatar
Mikaël Salson committed
145 146 147
  /**
   * Build a segmenter based on KmerSegmentation
   * @param seq: An object read from a FASTA/FASTQ file
148
   * @param multigermline: the multigermline
Mikaël Salson's avatar
Mikaël Salson committed
149
   */
150
  KmerSegmenter(Sequence seq, MultiGermline *multigermline);
Mathieu Giraud's avatar
Mathieu Giraud committed
151 152 153 154 155 156

  ~KmerSegmenter();

  /**
   * @return the KmerAffectAnalyser of the current sequence.
   */
157
  KmerAffectAnalyser *getKmerAffectAnalyser() const;
Mathieu Giraud's avatar
Mathieu Giraud committed
158 159 160 161 162 163 164 165 166 167

  /**
   * @return the status of the segmentation. Tells if the Sequence has been segmented
   *         of if it has not, what the reason is.
   * @assert getSegmentationStatus() == SEG_PLUS || getSegmentationStatus() == SEG_MINUS
   *         <==> isSegmented()
   */
  int getSegmentationStatus() const;

 private:
168
  void computeSegmentation(int strand, Germline* germline);
Mikaël Salson's avatar
Mikaël Salson committed
169 170 171 172 173
};

class FineSegmenter : public Segmenter
{
 public:
174
   int because;
Mikaël Salson's avatar
Mikaël Salson committed
175 176 177 178
   vector<pair<int, int> > score_V;
   vector<pair<int, int> > score_D;
   vector<pair<int, int> > score_J;
   
Mikaël Salson's avatar
Mikaël Salson committed
179 180 181
   /**
   * Build a fineSegmenter based on KmerSegmentation
   * @param seq: An object read from a FASTA/FASTQ file
182
   * @param germline: germline used
Mikaël Salson's avatar
Mikaël Salson committed
183
   */
184
  FineSegmenter(Sequence seq, Germline *germline, Cost segment_cost);
Mikaël Salson's avatar
Mikaël Salson committed
185 186 187
  
  /**
  * extend segmentation from VJ to VDJ
188
  * @param germline: germline used
Mikaël Salson's avatar
Mikaël Salson committed
189
  */
190
  void FineSegmentD(Germline *germline);
Mathieu Giraud's avatar
Mathieu Giraud committed
191

192
  JsonList toJsonList(Germline *germline);
Mikaël Salson's avatar
Mikaël Salson committed
193
  
Mikaël Salson's avatar
Mikaël Salson committed
194 195 196 197 198
};



#endif