geom.cpp

   1 /*
   2  *  geom.cpp
   3  *  Mothur
   4  *
   5  *  Created by Thomas Ryabin on 2/23/09.
   6  *  Copyright 2009 __MyCompanyName__. All rights reserved.
   7  *
   8  */
   9
  10 #include "geom.h"
  11
  12 /***********************************************************************/
  13
  14 double Geom::kEq(double k, double spec){
  15         return k/(1-k)*pow(1-k, spec)/(1-pow(1-k, spec));
  16 }
  17
  18 RAbundVector Geom::getRAbundVector(SAbundVector* rank){
  19                 vector <int> rData;
  20                 int mr = 1;
  21                 int nb = 0;
  22                 int ns = 0;
  23
  24                 for(int i = rank->size()-1; i > 0; i--)
  25                 {
  26                         int cur = rank->get(i);
  27                         if(mr == 1 && cur > 0)
  28                                 mr = i;
  29                         nb += cur;
  30                         ns += i*cur;
  31                         for(int j = 0; j < cur; j++)
  32                                 rData.push_back(i);
  33                 }
  34
  35                 RAbundVector rav = RAbundVector(rData, mr, nb, ns);
  36                 rav.setLabel(rank->getLabel());
  37                 return rav;
  38 }
  39
  40 /***********************************************************************************/
  41
  42 /***********************************************************************************/
  43 EstOutput Geom::getValues(SAbundVector* rank){
  44         try {
  45                 data.resize(2,0);
  46
  47                 rdata = getRAbundVector(rank);
  48                 int numInd = rdata.getNumSeqs();
  49                 int numSpec = rdata.getNumBins();
  50                 int min = rdata.get(rdata.size()-1);
  51                 double k = .5;
  52                 double step = .49999;
  53
  54                 while(fabs(min - numInd*kEq(k, (double)numSpec)) > .0001) //This uses a binary search to find the value of k.
  55                 {
  56                         if(numInd*kEq(k, numSpec) > min)
  57                                 k += step;
  58                         else
  59                                 k -= step;
  60                         step /= 2;
  61                 }
  62
  63                 double cK = 1/(1-pow(1-k, numSpec));
  64                 double sumExp = 0;
  65                 double sumObs = 0;
  66                 double maxDiff = 0;
  67
  68                 for(int i = 0; i < rdata.size(); i++)
  69                 {
  70                         sumObs += rdata.get(i);
  71                         sumExp += numInd*cK*k*pow(1-k, i);
  72                         double diff = fabs(sumObs-sumExp);
  73                         if(diff > maxDiff)
  74                                 maxDiff = diff;
  75                 }
  76
  77                 double DStatistic = maxDiff/numInd;
  78                 double critVal = 0;
  79                 /*cout << "Geom:\n";
  80                 cout << "D-Statistic = " << DStatistic << "\n";
  81                 cout << "Critical value for 95% confidence interval = ";*/
  82                 if(rdata.size() > 20)
  83                 {
  84                         critVal = .886/sqrt(rdata.size());
  85                 }
  86                 else
  87                 {
  88                         KOSTable table;
  89                         critVal = table.getConfLimit(numSpec);
  90                 }
  91                 /*cout << critVal << "\n";
  92                 cout << "If D-Statistic is less than the critical value then the data fits the Geometric Series model w/ 95% confidence.\n\n";*/
  93
  94                 data[0] = DStatistic;
  95                 data[1] = critVal;
  96
  97                 if (isnan(data[0]) || isinf(data[0])) { data[0] = 0; }
  98                 if (isnan(data[1]) || isinf(data[1])) { data[1] = 0; }
  99
 100                 return data;
 101         }
 102         catch(exception& e) {
 103                 cout << "Standard Error: " << e.what() << " has occurred in the NPShannon class Function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 104                 exit(1);
 105         }
 106         catch(...) {
 107                 cout << "An unknown error has occurred in the NPShannon class function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 108                 exit(1);
 109         }
 110 }
 111
 112 /***********************************************************************/
 113
 114
 115
 116
 117
 118
 119