decalc.h

   1 #ifndef DECALC_H
   2 #define DECALC_H
   3 /*
   4  *  decalc.h
   5  *  Mothur
   6  *
   7  *  Created by Sarah Westcott on 7/22/09.
   8  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   9  *
  10  */
  11
  12 #include "mothur.h"
  13 #include "sequence.hpp"
  14
  15 /***********************************************************************/
  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 //this structure is necessary to determine the sequence that contributed to the outliers when we remove them
  24 //this way we can remove all scores that are contributed by outlier sequences.
  25 struct quanMember {
  26         float score;
  27         int member1;
  28         int member2;
  29         quanMember (float s, int m, int n) : score(s), member1(m),  member2(n) {}
  30         quanMember() {}
  31
  32 };
  33
  34 //********************************************************************************************************************
  35 class DeCalculator {
  36
  37         public:
  38
  39                 DeCalculator() { m = MothurOut::getInstance(); }
  40                 ~DeCalculator() {};
  41
  42                 vector<Sequence> findClosest(Sequence, vector<Sequence*>&, vector<Sequence*>&, int, int);  //takes querySeq, a reference db, filteredRefDB, numWanted, minSim
  43                 Sequence* findClosest(Sequence*, vector<Sequence*>);
  44                 set<int> getPos() {  return h;  }
  45                 void setMask(string);
  46                 void setAlignmentLength(int l) {  alignLength = l;  }
  47                 void runMask(Sequence*);
  48                 void trimSeqs(Sequence*, Sequence*, map<int, int>&);
  49                 map<int, int> trimSeqs(Sequence&, vector<Sequence>&);
  50                 void removeObviousOutliers(vector< vector<float> >&, int);
  51                 vector<float> calcFreq(vector<Sequence*>, string);
  52                 vector<int> findWindows(Sequence*, int, int, int&, int);
  53                 vector<float> calcObserved(Sequence*, Sequence*, vector<int>, int);
  54                 vector<float>  calcExpected(vector<float>, float);
  55                 vector<float>  findQav(vector<int>, int, vector<float>);
  56                 float calcDE(vector<float>, vector<float>);
  57                 float calcDist(Sequence*, Sequence*, int, int);
  58                 float getCoef(vector<float>, vector<float>);
  59                 vector< vector<float> > getQuantiles(vector<Sequence*>, vector<int>, int, vector<float>, int, int, int);
  60
  61                 vector<int> returnObviousOutliers(vector< vector<quanMember> >, int);
  62
  63                 map<int, int> getMaskMap() { return maskMap; }
  64
  65         private:
  66                 //vector<quanMember> sortContrib(map<quanMember*, float>);  //used by mallard
  67                 float findAverage(vector<float>);
  68                 //int findLargestContrib(vector<int>);
  69                 //void removeContrib(int, vector<quanMember>&);
  70                 string seqMask;
  71                 set<int> h;
  72                 int alignLength;
  73                 map<int, int> maskMap;
  74                 MothurOut* m;
  75
  76 };
  77
  78 /***********************************************************************/
  79
  80 #endif