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
  33
  34         private:
  35
  36                 struct linePair {
  37                         int start;
  38                         int end;
  39                         linePair(int i, int j) : start(i), end(j) {}
  40                 };
  41
  42                 Dist* distcalculator;
  43                 int iters;
  44                 string fastafile, templateFile, consfile;
  45                 vector<linePair*> lines;
  46                 vector<Sequence*> querySeqs;
  47                 vector<Sequence*> templateSeqs;
  48
  49                 vector<Sequence> bestfit;  //bestfit[0] matches queryseqs[0]...
  50
  51                 vector< vector<float> > obsDistance;  //obsDistance[0] is the vector of observed distances for queryseqs[0]...
  52                 vector< vector<float> > expectedDistance;  //expectedDistance[0] is the vector of expected distances for queryseqs[0]...
  53                 vector<float> deviation;  //deviation[0] is the percentage of mismatched pairs over the whole seq between querySeqs[0] and its best match.
  54                 vector< vector<int> > windows;  // windows[0] is a vector containing the starting spot in queryseqs[0] aligned sequence for each window.
  55                                                                                 //this is needed so you can move by bases and not just spots in the alignment
  56                 vector< map<int, int> >  trim;  //trim[0] is the start and end position of trimmed querySeqs[0].  Used to find the variability over each sequence window.
  57
  58                 vector<int> windowSizes;    //windowSizes[0] = window size of querySeqs[0]
  59
  60                 vector< vector<float> > Qav;    //Qav[0] is the vector of average variablility for queryseqs[0]...
  61                 vector<float>  seqCoef;                         //seqCoef[0] is the coeff for queryseqs[0]...
  62                 vector<float> DE;                                       //DE[0] is the deviaation for queryseqs[0]...
  63                 vector<float> probabilityProfile;
  64
  65                 vector<Sequence*> readSeqs(string);
  66                 void trimSeqs(Sequence*, Sequence&, int);
  67                 vector<float> readFreq();
  68                 vector< vector<float> > findQav(int, int);
  69                 vector<float> calcFreq(vector<Sequence*>);
  70                 vector<float> getCoef(int, int);
  71
  72                 vector<Sequence> findPairs(int, int);
  73                 vector< vector<int> > findWindows(int, int);
  74                 vector< vector<float> > calcObserved(int, int);
  75                 vector< vector<float> > calcExpected(int, int);
  76                 vector<float> calcDE(int, int);
  77                 vector<float> calcDist(int, int);
  78
  79                 void createProcessesSpots();
  80                 void createProcesses();
  81
  82
  83
  84 };
  85
  86 /***********************************************************/
  87
  88 #endif
  89