shhhercommand.h

   1 #ifndef SHHHER_H
   2 #define SHHHER_H
   3
   4 /*
   5  *  shhher.h
   6  *  Mothur
   7  *
   8  *  Created by Pat Schloss on 12/27/10.
   9  *  Copyright 2010 Schloss Lab. All rights reserved.
  10  *
  11  */
  12
  13 #include "mothur.h"
  14 #include "command.hpp"
  15
  16
  17 class ShhherCommand : public Command {
  18
  19 public:
  20         ShhherCommand(string);
  21         ShhherCommand();
  22         ~ShhherCommand() {}
  23
  24         vector<string> setParameters();
  25         string getCommandName()                 { return "shhh.seqs";   }
  26         string getCommandCategory()             { return "Hidden";              }
  27         string getHelpString();
  28         string getCitation() { return "no citation"; }
  29
  30         int execute();
  31         void help() { m->mothurOut(getHelpString()); }
  32 private:
  33
  34         int abort;
  35
  36         string outputDir, flowFileName, flowFilesFileName, lookupFileName, compositeFASTAFileName;
  37
  38         int processors, maxIters;
  39         float cutoff, sigma, minDelta;
  40         string flowOrder;
  41
  42         vector<int> nSeqsBreaks;
  43         vector<int> nOTUsBreaks;
  44         vector<double> singleLookUp;
  45         vector<double> jointLookUp;
  46
  47         vector<string> seqNameVector;
  48         vector<int> lengths;
  49         vector<short> flowDataIntI;
  50         vector<double> flowDataPrI;
  51         map<string, int> nameMap;
  52         vector<int> otuData;
  53         vector<int> cumNumSeqs;
  54         vector<int> nSeqsPerOTU;
  55         vector<vector<int> > aaP;       //tMaster->aanP:        each row is a different otu / each col contains the sequence indices
  56         vector<vector<int> > aaI;       //tMaster->aanI:        that are in each otu - can't differentiate between aaP and aaI
  57         vector<int> seqNumber;          //tMaster->anP:         the sequence id number sorted by OTU
  58         vector<int> seqIndex;           //tMaster->anI;         the index that corresponds to seqNumber
  59         vector<double> dist;            //adDist - distance of sequences to centroids
  60         vector<short> change;           //did the centroid sequence change? 0 = no; 1 = yes
  61         vector<int> centroids;          //the representative flowgram for each cluster m
  62         vector<double> weight;
  63         vector<double> singleTau;       //tMaster->adTau:       1-D Tau vector (1xnumSeqs)
  64         vector<short> uniqueFlowgrams;
  65         vector<int> uniqueCount;
  66         vector<int> mapSeqToUnique;
  67         vector<int> mapUniqueToSeq;
  68         vector<int> uniqueLengths;
  69
  70         int numSeqs, numUniques, numOTUs, numFlowCells;
  71
  72         void getSingleLookUp();
  73         void getJointLookUp();
  74         void getFlowData();
  75         void getUniques();
  76         double getProbIntensity(int);
  77         float calcPairwiseDist(int, int);
  78         void flowDistParentFork(string, int, int);
  79
  80         string createDistFile(int);
  81         string createNamesFile();
  82         string cluster(string, string);
  83
  84         void getOTUData(string);
  85         void initPyroCluster();
  86         void fill();
  87         void calcCentroids();
  88         void calcCentroidsDriver(int, int);
  89         double getDistToCentroid(int, int, int);
  90         double getNewWeights();
  91         double getLikelihood();
  92         void checkCentroids();
  93         void calcNewDistances();
  94         void calcNewDistancesParent(int, int);
  95         void calcNewDistancesChild(int, int, vector<int>&, vector<int>&, vector<double>&);
  96
  97
  98         void setOTUs();
  99         void writeQualities(vector<int>);
 100         void writeSequences(vector<int>);
 101         void writeNames(vector<int>);
 102         void writeGroups();
 103         void writeClusters(vector<int>);
 104
 105
 106 #ifdef USE_MPI
 107         string flowDistMPI(int, int);
 108         void calcNewDistancesChildMPI(int, int, vector<int>&);
 109
 110         int pid, ncpus;
 111 #endif
 112
 113 };
 114
 115
 116 #endif
 117