segment.h 6.58 KB
Newer Older
Mikaël Salson's avatar
Mikaël Salson committed
1 2 3 4
#ifndef SEGMENT_H
#define SEGMENT_H

#include <string>
5
#include <fstream>
Mikaël Salson's avatar
Mikaël Salson committed
6 7 8 9
#include <iostream>
#include "fasta.h"
#include "dynprog.h"
#include "tools.h"
10
#include "germline.h"
Mikaël Salson's avatar
Mikaël Salson committed
11 12
#include "kmerstore.h"
#include "kmeraffect.h"
Mathieu Giraud's avatar
Mathieu Giraud committed
13
#include "affectanalyser.h"
Marc Duez's avatar
Marc Duez committed
14
#include "../lib/json.hpp"
Mikaël Salson's avatar
Mikaël Salson committed
15 16

#define EXTEND_D_ZONE 5
17 18 19

#define MIN_D_LENGTH 5          /* If a D-REGION is smaller than this threshold, it is not output */

20
#define RATIO_STRAND 2          /* The ratio between the affectations in one
Mikaël Salson's avatar
Mikaël Salson committed
21 22
                                   strand and the other, to safely attribute a
                                   segment to a given strand */
Mikaël Salson's avatar
Mikaël Salson committed
23

24 25 26 27
#define DETECT_THRESHOLD_STRAND 5   /* If the number of total affectations
                                       is above this threshold, then a sequence with no clearly attributed
                                       stranf will be marked as STRAND_NOT_CONSISTEN */

28
#define DETECT_THRESHOLD 5      /* If the number of both V and J affectations
29 30
                                   is above this threshold, then the sequence,
                                   if it not segmented, will be marked as AMBIGUOUS */
31

Mathieu Giraud's avatar
Mathieu Giraud committed
32 33
#define JSON_REMEMBER_BEST  4   /* The number of V/D/J predictions to keep  */

34 35
#define NO_LIMIT_VALUE  -1

36
#define THRESHOLD_NB_EXPECTED 1.0 /* Threshold of the accepted expected value for number of found k-mers */
37 38


39

Mikaël Salson's avatar
Mikaël Salson committed
40
using namespace std;
Marc Duez's avatar
Marc Duez committed
41
using json = nlohmann::json;
Mikaël Salson's avatar
Mikaël Salson committed
42

43 44 45 46
enum SEGMENTED { NOT_PROCESSED,
		 TOTAL_SEG_AND_WINDOW,
                 SEG_PLUS, SEG_MINUS,
                 UNSEG_TOO_SHORT, UNSEG_STRAND_NOT_CONSISTENT,
Mikaël Salson's avatar
Mikaël Salson committed
47
		 UNSEG_TOO_FEW_ZERO,  UNSEG_TOO_FEW_V, UNSEG_TOO_FEW_J, 
48
                 UNSEG_BAD_DELTA_MIN, UNSEG_AMBIGUOUS,
49
		 UNSEG_TOO_SHORT_FOR_WINDOW,
50

Mikaël Salson's avatar
Mikaël Salson committed
51
		 STATS_SIZE } ;
52 53 54 55
const char* const segmented_mesg[] = { "?",
                                       "SEG",
                                       "SEG_+", "SEG_-",
                                       "UNSEG too short", "UNSEG strand",
56
				       "UNSEG too few (0)", "UNSEG too few V", "UNSEG too few J",
57
				       "UNSEG < delta_min", "UNSEG ambiguous",
58
                                       "UNSEG too short w",
Mikaël Salson's avatar
Mikaël Salson committed
59
                                      } ;
Mikaël Salson's avatar
Mikaël Salson committed
60 61 62 63

class Segmenter {
protected:
  string sequence;
Mathieu Giraud's avatar
Mathieu Giraud committed
64 65
  int Vend, Jstart;
  int Dstart, Dend;
Marc Duez's avatar
Marc Duez committed
66
  int CDR3start, CDR3end;
Mikaël Salson's avatar
Mikaël Salson committed
67
  bool reversed, segmented, dSegmented;
68
  int because;
Mikaël Salson's avatar
Mikaël Salson committed
69 70 71 72 73 74

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

 public:
75
  Germline *segmented_germline;
76
  string label;
Mikaël Salson's avatar
Mikaël Salson committed
77
  string code;
Mikaël Salson's avatar
Mikaël Salson committed
78
  string code_short;
Mikaël Salson's avatar
Mikaël Salson committed
79
  string code_light;
80 81
  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
82 83
  int best_V, best_J ;
  int del_V, del_D_left, del_D_right, del_J ;
84
  string seg_V, seg_N, seg_J, system;
85

Mikaël Salson's avatar
Mikaël Salson committed
86
  int best_D;
87
  double evalue;
88 89
  double evalue_left;
  double evalue_right;
Mikaël Salson's avatar
Mikaël Salson committed
90 91 92
  string seg_N1, seg_D, seg_N2;
  Cost segment_cost;

Mathieu Giraud's avatar
Mathieu Giraud committed
93
  virtual ~Segmenter();
Mikaël Salson's avatar
Mikaël Salson committed
94

Mathieu Giraud's avatar
Mathieu Giraud committed
95
  /* Queries */
Mikaël Salson's avatar
Mikaël Salson committed
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146



  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;

147 148 149 150 151 152 153 154 155
  /**
   * @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;

  string getInfoLine() const;
Mikaël Salson's avatar
Mikaël Salson committed
156

157 158 159 160 161
  /**
   * @post status == SEG_PLUS || status == SEG_MINUS <==> isSegmented()
   */
  void setSegmentationStatus(int status);

Mikaël Salson's avatar
Mikaël Salson committed
162 163 164 165 166 167 168 169 170 171 172
  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
173
 private:
174
  KmerAffectAnalyser *kaa;
Mikaël Salson's avatar
Mikaël Salson committed
175 176 177 178
 protected:
  string affects;

 public:
179
  int score;
180 181
  int pvalue_left;
  int pvalue_right;
182
  KmerAffect before, after;
183 184

  KmerSegmenter();
Mikaël Salson's avatar
Mikaël Salson committed
185 186 187
  /**
   * Build a segmenter based on KmerSegmentation
   * @param seq: An object read from a FASTA/FASTQ file
188
   * @param germline: the germline
Mikaël Salson's avatar
Mikaël Salson committed
189
   */
190
  KmerSegmenter(Sequence seq, Germline *germline, double threshold = THRESHOLD_NB_EXPECTED, int multiplier=1);
Mathieu Giraud's avatar
Mathieu Giraud committed
191

192 193
  KmerSegmenter(const KmerSegmenter &seg);

Mathieu Giraud's avatar
Mathieu Giraud committed
194 195 196 197 198
  ~KmerSegmenter();

  /**
   * @return the KmerAffectAnalyser of the current sequence.
   */
199
  KmerAffectAnalyser *getKmerAffectAnalyser() const;
Mathieu Giraud's avatar
Mathieu Giraud committed
200

201
  string getInfoLineWithAffects() const;
Marc Duez's avatar
Marc Duez committed
202
  json toJson();
203

Mathieu Giraud's avatar
Mathieu Giraud committed
204
 private:
205 206
  void computeSegmentation(int strand, KmerAffect left, KmerAffect right,
                           double threshold, int multiplier);
Mikaël Salson's avatar
Mikaël Salson committed
207 208
};

209 210 211

class KmerMultiSegmenter
{
212 213
 private:
  double threshold_nb_expected;
214 215 216
 public:
  /**
   * @param seq: An object read from a FASTA/FASTQ file
217 218 219 220
   * @param multigermline: the multigerm
   * @param threshold: threshold of randomly expected segmentation
   */
  KmerMultiSegmenter(Sequence seq, MultiGermline *multigermline, ostream *out_unsegmented,
221
                     double threshold = THRESHOLD_NB_EXPECTED, int nb_reads_for_evalue = 1);
222

223 224
  ~KmerMultiSegmenter();

225
  KmerSegmenter *the_kseg;
226
  MultiGermline *multi_germline;
227 228 229
};


Mikaël Salson's avatar
Mikaël Salson committed
230 231 232
class FineSegmenter : public Segmenter
{
 public:
Mikaël Salson's avatar
Mikaël Salson committed
233 234 235 236
   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
237 238 239
   /**
   * Build a fineSegmenter based on KmerSegmentation
   * @param seq: An object read from a FASTA/FASTQ file
240
   * @param germline: germline used
Mikaël Salson's avatar
Mikaël Salson committed
241
   */
242
  FineSegmenter(Sequence seq, Germline *germline, Cost segment_cost);
Mikaël Salson's avatar
Mikaël Salson committed
243 244 245
  
  /**
  * extend segmentation from VJ to VDJ
246
  * @param germline: germline used
Mikaël Salson's avatar
Mikaël Salson committed
247
  */
248
  void FineSegmentD(Germline *germline);
Marc Duez's avatar
Marc Duez committed
249
  void findCDR3();
Mathieu Giraud's avatar
Mathieu Giraud committed
250

Marc Duez's avatar
Marc Duez committed
251
  json toJson();
Mikaël Salson's avatar
Mikaël Salson committed
252
  
Mikaël Salson's avatar
Mikaël Salson committed
253 254 255 256 257
};



#endif