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
  16 //This class was created using the algorythms described in the
  17 // "At Least 1 in 20 16S rRNA Sequence Records Currently Held in the Public Repositories is Estimated To Contain Substantial Anomalies" paper
  18 //by Kevin E. Ashelford 1, Nadia A. Chuzhanova 3, John C. Fry 1, Antonia J. Jones 2 and Andrew J. Weightman 1.
  19
  20 /***********************************************************/
  21
  22 class Pintail : public Chimera {
  23
  24         public:
  25                 Pintail(string, string);
  26                 ~Pintail();
  27
  28                 void getChimeras();
  29                 void print(ostream&);
  30
  31                 void setCons(string c)          { consfile = c;  }
  32                 void setQuantiles(string q) { quanfile = q;  }
  33
  34
  35         private:
  36
  37                 struct linePair {
  38                         int start;
  39                         int end;
  40                         linePair(int i, int j) : start(i), end(j) {}
  41                         linePair(){}
  42                 };
  43
  44                 Dist* distcalculator;
  45                 int iters;
  46                 string fastafile, templateFile, consfile, quanfile;
  47
  48                 vector<linePair*> lines;
  49                 vector<linePair*> templateLines;
  50                 vector<Sequence*> querySeqs;
  51                 vector<Sequence*> templateSeqs;
  52
  53                 vector<Sequence> bestfit;  //bestfit[0] matches queryseqs[0]...
  54
  55                 vector< vector<float> > obsDistance;  //obsDistance[0] is the vector of observed distances for queryseqs[0]...
  56                 vector< vector<float> > expectedDistance;  //expectedDistance[0] is the vector of expected distances for queryseqs[0]...
  57                 vector<float> deviation;  //deviation[0] is the percentage of mismatched pairs over the whole seq between querySeqs[0] and its best match.
  58                 vector< vector<int> > windowsForeachQuery;  // windowsForeachQuery[0] is a vector containing the starting spot in queryseqs[0] aligned sequence for each window.
  59                                                                                 //this is needed so you can move by bases and not just spots in the alignment
  60
  61                 vector<int> windowSizes;                        //windowSizes[0] = window size of querySeqs[0]
  62                 vector<int> windowSizesTemplate;    //windowSizesTemplate[0] = window size of templateSeqs[0]
  63
  64                 vector< map<int, int> > trimmed;    //trimmed[0] = start and stop of trimmed sequences for querySeqs[0]
  65                 map<int, int>::iterator it;
  66
  67                 vector< vector<float> > Qav;    //Qav[0] is the vector of average variablility for queryseqs[0]...
  68                 vector<float>  seqCoef;                         //seqCoef[0] is the coeff for queryseqs[0]...
  69                 vector<float> DE;                                       //DE[0] is the deviaation for queryseqs[0]...
  70                 vector<float> probabilityProfile;
  71                 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...
  72
  73                 void trimSeqs(Sequence*, Sequence, map<int, int>&);
  74                 vector<float> readFreq();
  75                 vector< vector<float> > readQuantiles();
  76                 vector< vector<float> > getQuantiles(int, int);
  77                 vector<float> calcFreq(vector<Sequence*>);
  78
  79                 vector<Sequence> findPairs(int, int);
  80                 vector<int> findWindows(Sequence*, int, int, int&);
  81                 vector<float> calcObserved(Sequence*, Sequence, vector<int>, int);
  82                 vector<float>  calcExpected(vector<float>, float);
  83                 vector<float>  findQav(vector<int>, int);
  84                 float calcDE(vector<float>, vector<float>);
  85                 float calcDist(Sequence*, Sequence, int, int);
  86                 float getCoef(vector<float>, vector<float>);
  87
  88                 void createProcessesSpots();
  89                 void createProcesses();
  90                 void createProcessesQuan();
  91
  92
  93
  94 };
  95
  96 /***********************************************************/
  97
  98 #endif
  99