Mise à jour terminée. Pour connaître les apports de la version 13.8.4 par rapport à notre ancienne version vous pouvez lire les "Release Notes" suivantes :
https://about.gitlab.com/releases/2021/02/11/security-release-gitlab-13-8-4-released/
https://about.gitlab.com/releases/2021/02/05/gitlab-13-8-3-released/

segment.h 6.29 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
16 17 18

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

Mikaël Salson's avatar
Mikaël Salson committed
19 20 21
#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
22

23
#define DETECT_THRESHOLD 5      /* If the number of both V and J affectations
24 25 26 27
                                   is above this threshold, then the sequence
                                   will be labeled as 'detected', and, if it
                                   not segmented, the remaining germlines will
                                   not be tested */
28

Mathieu Giraud's avatar
Mathieu Giraud committed
29 30
#define JSON_REMEMBER_BEST  4   /* The number of V/D/J predictions to keep  */

31 32
#define NO_LIMIT_VALUE  -1

33
#define THRESHOLD_NB_EXPECTED 1.0 /* Threshold of the accepted expected value for number of found k-mers */
34 35


36

Mikaël Salson's avatar
Mikaël Salson committed
37 38 39 40
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, 
41
		 UNSEG_BAD_DELTA_MIN, UNSEG_BAD_DELTA_MAX, UNSEG_AMBIGUOUS, UNSEG_NOISY,
42
		 UNSEG_TOO_SHORT_FOR_WINDOW,
Mikaël Salson's avatar
Mikaël Salson committed
43 44 45
		 TOTAL_SEG_AND_WINDOW, 
		 STATS_SIZE } ;
const char* const segmented_mesg[] = { "?", "SEG_+", "SEG_-", "UNSEG too short", "UNSEG strand",  
46
				       "UNSEG too few (0)", "UNSEG too few V", "UNSEG too few J",
47
				       "UNSEG < delta_min", "UNSEG > delta_max", "UNSEG ambiguous",
48
                                       "UNSEG noisy",
49 50
                                       "UNSEG too short w",
                                       "SEG",
Mikaël Salson's avatar
Mikaël Salson committed
51
                                      } ;
Mikaël Salson's avatar
Mikaël Salson committed
52 53 54 55 56

class Segmenter {
protected:
  string label;
  string sequence;
Mathieu Giraud's avatar
Mathieu Giraud committed
57 58
  int Vend, Jstart;
  int Dstart, Dend;
Marc Duez's avatar
Marc Duez committed
59
  int CDR3start, CDR3end;
Mikaël Salson's avatar
Mikaël Salson committed
60
  bool reversed, segmented, dSegmented;
61
  int because;
Mikaël Salson's avatar
Mikaël Salson committed
62 63 64 65 66 67

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

 public:
68
  Germline *segmented_germline;
Mikaël Salson's avatar
Mikaël Salson committed
69
  string code;
Mikaël Salson's avatar
Mikaël Salson committed
70
  string code_short;
Mikaël Salson's avatar
Mikaël Salson committed
71
  string code_light;
72 73
  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
74 75
  int best_V, best_J ;
  int del_V, del_D_left, del_D_right, del_J ;
76
  string seg_V, seg_N, seg_J, system;
77

Mikaël Salson's avatar
Mikaël Salson committed
78
  int best_D;
79
  double evalue;
80 81
  double evalue_left;
  double evalue_right;
Mikaël Salson's avatar
Mikaël Salson committed
82 83 84
  string seg_N1, seg_D, seg_N2;
  Cost segment_cost;

Mathieu Giraud's avatar
Mathieu Giraud committed
85
  virtual ~Segmenter();
Mikaël Salson's avatar
Mikaël Salson committed
86

Mathieu Giraud's avatar
Mathieu Giraud committed
87
  /* Queries */
Mikaël Salson's avatar
Mikaël Salson committed
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138



  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;

139 140 141 142 143 144 145 146 147
  /**
   * @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
148

149 150 151 152 153
  /**
   * @post status == SEG_PLUS || status == SEG_MINUS <==> isSegmented()
   */
  void setSegmentationStatus(int status);

Mikaël Salson's avatar
Mikaël Salson committed
154 155 156 157 158 159 160 161 162 163 164
  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
165
 private:
166
  int detected;
167
  KmerAffectAnalyser *kaa;
Mikaël Salson's avatar
Mikaël Salson committed
168 169 170 171
 protected:
  string affects;

 public:
172
  bool isDetected() const;
173
  int score;
174 175
  int pvalue_left;
  int pvalue_right;
176 177

  KmerSegmenter();
Mikaël Salson's avatar
Mikaël Salson committed
178 179 180
  /**
   * Build a segmenter based on KmerSegmentation
   * @param seq: An object read from a FASTA/FASTQ file
181
   * @param germline: the germline
Mikaël Salson's avatar
Mikaël Salson committed
182
   */
183
  KmerSegmenter(Sequence seq, Germline *germline, int multiplier=1);
Mathieu Giraud's avatar
Mathieu Giraud committed
184

185 186
  KmerSegmenter(const KmerSegmenter &seg);

Mathieu Giraud's avatar
Mathieu Giraud committed
187 188 189 190 191
  ~KmerSegmenter();

  /**
   * @return the KmerAffectAnalyser of the current sequence.
   */
192
  KmerAffectAnalyser *getKmerAffectAnalyser() const;
Mathieu Giraud's avatar
Mathieu Giraud committed
193

194 195
  void toJsonList(JsonList *seg);

Mathieu Giraud's avatar
Mathieu Giraud committed
196
 private:
197
  void computeSegmentation(int strand, KmerAffect left, KmerAffect right, int multiplier);
Mikaël Salson's avatar
Mikaël Salson committed
198 199
};

200 201 202

class KmerMultiSegmenter
{
203 204
 private:
  double threshold_nb_expected;
205 206 207
 public:
  /**
   * @param seq: An object read from a FASTA/FASTQ file
208 209 210 211
   * @param multigermline: the multigerm
   * @param threshold: threshold of randomly expected segmentation
   */
  KmerMultiSegmenter(Sequence seq, MultiGermline *multigermline, ostream *out_unsegmented,
212
                     double threshold = THRESHOLD_NB_EXPECTED, int nb_reads_for_evalue = 1);
213

214 215
  ~KmerMultiSegmenter();

216
  KmerSegmenter *the_kseg;
217
  MultiGermline *multi_germline;
218 219 220
};


Mikaël Salson's avatar
Mikaël Salson committed
221 222 223
class FineSegmenter : public Segmenter
{
 public:
Mikaël Salson's avatar
Mikaël Salson committed
224 225 226 227
   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
228 229 230
   /**
   * Build a fineSegmenter based on KmerSegmentation
   * @param seq: An object read from a FASTA/FASTQ file
231
   * @param germline: germline used
Mikaël Salson's avatar
Mikaël Salson committed
232
   */
233
  FineSegmenter(Sequence seq, Germline *germline, Cost segment_cost);
Mikaël Salson's avatar
Mikaël Salson committed
234 235 236
  
  /**
  * extend segmentation from VJ to VDJ
237
  * @param germline: germline used
Mikaël Salson's avatar
Mikaël Salson committed
238
  */
239
  void FineSegmentD(Germline *germline);
Marc Duez's avatar
Marc Duez committed
240
  void findCDR3();
Mathieu Giraud's avatar
Mathieu Giraud committed
241

242
  void toJsonList(JsonList *seg);
Mikaël Salson's avatar
Mikaël Salson committed
243
  
Mikaël Salson's avatar
Mikaël Salson committed
244 245 246 247 248
};



#endif