subsample.cpp

   1 //
   2 //  subsample.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 4/2/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "subsample.h"
  10
  11 //**********************************************************************************************************************
  12 vector<string> SubSample::getSample(vector<SharedRAbundVector*>& thislookup, int size) {
  13         try {
  14
  15                 //save mothurOut's binLabels to restore for next label
  16                 vector<string> saveBinLabels = m->currentBinLabels;
  17
  18                 int numBins = thislookup[0]->getNumBins();
  19                 for (int i = 0; i < thislookup.size(); i++) {
  20                         int thisSize = thislookup[i]->getNumSeqs();
  21
  22                         if (thisSize != size) {
  23
  24                                 string thisgroup = thislookup[i]->getGroup();
  25
  26                                 OrderVector order;
  27                                 for(int p=0;p<numBins;p++){
  28                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
  29                                                 order.push_back(p);
  30                                         }
  31                                 }
  32                                 random_shuffle(order.begin(), order.end());
  33
  34                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
  35                                 temp->setLabel(thislookup[i]->getLabel());
  36                                 temp->setGroup(thislookup[i]->getGroup());
  37
  38                                 delete thislookup[i];
  39                                 thislookup[i] = temp;
  40
  41
  42                                 for (int j = 0; j < size; j++) {
  43
  44                                         if (m->control_pressed) {  return m->currentBinLabels; }
  45
  46                                         int bin = order.get(j);
  47
  48                                         int abund = thislookup[i]->getAbundance(bin);
  49                                         thislookup[i]->set(bin, (abund+1), thisgroup);
  50                                 }
  51                         }
  52                 }
  53
  54                 //subsampling may have created some otus with no sequences in them
  55                 eliminateZeroOTUS(thislookup);
  56
  57                 if (m->control_pressed) { return m->currentBinLabels; }
  58
  59                 //save mothurOut's binLabels to restore for next label
  60         vector<string> subsampleBinLabels = m->currentBinLabels;
  61                 m->currentBinLabels = saveBinLabels;
  62
  63                 return subsampleBinLabels;
  64
  65         }
  66         catch(exception& e) {
  67                 m->errorOut(e, "SubSample", "getSample");
  68                 exit(1);
  69         }
  70 }
  71 //**********************************************************************************************************************
  72 int SubSample::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
  73         try {
  74
  75                 vector<SharedRAbundVector*> newLookup;
  76                 for (int i = 0; i < thislookup.size(); i++) {
  77                         SharedRAbundVector* temp = new SharedRAbundVector();
  78                         temp->setLabel(thislookup[i]->getLabel());
  79                         temp->setGroup(thislookup[i]->getGroup());
  80                         newLookup.push_back(temp);
  81                 }
  82
  83                 //for each bin
  84                 vector<string> newBinLabels;
  85                 string snumBins = toString(thislookup[0]->getNumBins());
  86                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
  87                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
  88
  89                         //look at each sharedRabund and make sure they are not all zero
  90                         bool allZero = true;
  91                         for (int j = 0; j < thislookup.size(); j++) {
  92                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
  93                         }
  94
  95                         //if they are not all zero add this bin
  96                         if (!allZero) {
  97                                 for (int j = 0; j < thislookup.size(); j++) {
  98                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
  99                                 }
 100                                 //if there is a bin label use it otherwise make one
 101                                 string binLabel = "Otu";
 102                                 string sbinNumber = toString(i+1);
 103                                 if (sbinNumber.length() < snumBins.length()) {
 104                                         int diff = snumBins.length() - sbinNumber.length();
 105                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
 106                                 }
 107                                 binLabel += sbinNumber;
 108                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
 109
 110                                 newBinLabels.push_back(binLabel);
 111                         }
 112                 }
 113
 114                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
 115                 thislookup.clear();
 116
 117                 thislookup = newLookup;
 118                 m->currentBinLabels = newBinLabels;
 119
 120                 return 0;
 121
 122         }
 123         catch(exception& e) {
 124                 m->errorOut(e, "SubSample", "eliminateZeroOTUS");
 125                 exit(1);
 126         }
 127 }
 128
 129
 130 //**********************************************************************************************************************
 131
 132