pintail.h

   1 #ifndef PINTAIL_H
   2 #define PINTAIL_H
   3
   4 /*
   5  *  pintail.h
   6  *  Mothur
   7  *
   8  *  Created by Sarah Westcott on 7/9/09.
   9  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  10  *
  11  */
  12
  13 #include "chimera.h"
  14 #include "dist.h"
  15 #include "decalc.h"
  16
  17 //This class was created using the algorythms described in the
  18 // "At Least 1 in 20 16S rRNA Sequence Records Currently Held in the Public Repositories is Estimated To Contain Substantial Anomalies" paper
  19 //by Kevin E. Ashelford 1, Nadia A. Chuzhanova 3, John C. Fry 1, Antonia J. Jones 2 and Andrew J. Weightman 1.
  20
  21 /***********************************************************/
  22
  23 class Pintail : public Chimera {
  24
  25         public:
  26                 Pintail(string, string);
  27                 ~Pintail();
  28
  29                 void getChimeras();
  30                 void print(ostream&);
  31
  32                 void setCons(string c)          { consfile = c;  }
  33                 void setQuantiles(string q) { quanfile = q;  }
  34
  35
  36         private:
  37
  38                 struct linePair {
  39                         int start;
  40                         int end;
  41                         linePair(int i, int j) : start(i), end(j) {}
  42                         linePair(){}
  43                 };
  44
  45                 Dist* distcalculator;
  46                 DeCalculator* decalc;
  47                 int iters;
  48                 string fastafile, templateFile, consfile, quanfile;
  49
  50
  51                 vector<linePair*> lines;
  52                 vector<linePair*> templateLines;
  53                 vector<Sequence*> querySeqs;
  54                 vector<Sequence*> templateSeqs;
  55
  56                 vector<Sequence*> bestfit;  //bestfit[0] matches queryseqs[0]...
  57
  58                 vector< vector<float> > obsDistance;  //obsDistance[0] is the vector of observed distances for queryseqs[0]...
  59                 vector< vector<float> > expectedDistance;  //expectedDistance[0] is the vector of expected distances for queryseqs[0]...
  60                 vector<float> deviation;  //deviation[0] is the percentage of mismatched pairs over the whole seq between querySeqs[0] and its best match.
  61                 vector< vector<int> > windowsForeachQuery;  // windowsForeachQuery[0] is a vector containing the starting spot in queryseqs[0] aligned sequence for each window.
  62                                                                                 //this is needed so you can move by bases and not just spots in the alignment
  63
  64                 vector<int> windowSizes;                        //windowSizes[0] = window size of querySeqs[0]
  65                 vector<int> windowSizesTemplate;    //windowSizesTemplate[0] = window size of templateSeqs[0]
  66
  67                 vector< map<int, int> > trimmed;    //trimmed[0] = start and stop of trimmed sequences for querySeqs[0]
  68                 map<int, int>::iterator it;
  69
  70                 vector< vector<float> > Qav;    //Qav[0] is the vector of average variablility for queryseqs[0]...
  71                 vector<float>  seqCoef;                         //seqCoef[0] is the coeff for queryseqs[0]...
  72                 vector<float> DE;                                       //DE[0] is the deviaation for queryseqs[0]...
  73                 vector<float> probabilityProfile;
  74                 vector< vector<float> > quantiles;  //quantiles[0] is the vector of deviations with ceiling score of 1, quantiles[1] is the vector of deviations with ceiling score of 2...
  75                 vector< vector<quanMember> > quantilesMembers;  //quantiles[0] is the vector of deviations with ceiling score of 1, quantiles[1] is the vector of deviations with ceiling score of 2...
  76                 vector< set<int> > h;
  77
  78
  79                 vector<float> readFreq();
  80                 vector< vector<float> > readQuantiles();
  81                 vector<Sequence*> findPairs(int, int);
  82
  83                 void createProcessesSpots();
  84                 void createProcessesPairs();
  85                 void createProcesses();
  86                 void createProcessesQuan();
  87
  88                 vector<float> makeCompliant;  //used by decalc->getQuantiles so pintail and mallard can use same function, it contains the highest de value for each seq in the template
  89
  90 };
  91
  92 /***********************************************************/
  93
  94 #endif
  95