getoturepcommand.h

   1 #ifndef GETOTUREPCOMMAND_H
   2 #define GETOTUREPCOMMAND_H
   3 /*
   4  *  getoturepcommand.h
   5  *  Mothur
   6  *
   7  *  Created by Sarah Westcott on 4/6/09.
   8  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   9  *
  10  */
  11
  12  /* The get.oturep command outputs a .fastarep file for each distance you specify, selecting one OTU representative for each bin. */
  13
  14 #include "command.hpp"
  15 #include "globaldata.hpp"
  16 #include "listvector.hpp"
  17 #include "inputdata.h"
  18 #include "readotu.h"
  19 #include "fastamap.h"
  20 #include "groupmap.h"
  21 #include "readmatrix.hpp"
  22 #include "formatmatrix.h"
  23
  24 typedef list<PCell>::iterator MatData;
  25 typedef map<int, float> SeqMap;
  26
  27 struct repStruct {
  28                 string name;
  29                 int     bin;
  30                 int size;
  31                 string group;
  32
  33                 repStruct(){}
  34                 repStruct(string n, int b, int s, string g) : name(n), bin(b), size(s), group(g) {}
  35                 ~repStruct() {}
  36 };
  37
  38 class GetOTURepCommand : public Command {
  39
  40 public:
  41         GetOTURepCommand(string);
  42         GetOTURepCommand();
  43         ~GetOTURepCommand();
  44         vector<string> getRequiredParameters();
  45         vector<string> getValidParameters();
  46         vector<string> getRequiredFiles();
  47         map<string, vector<string> > getOutputFiles() { return outputTypes; }
  48         int execute();
  49         void help();
  50
  51 private:
  52         GlobalData* globaldata;
  53         ListVector* list;
  54         ReadOTUFile* read;
  55         InputData* input;
  56         FastaMap* fasta;
  57         GroupMap* groupMap;
  58         ReadMatrix* readMatrix;
  59         FormatMatrix* formatMatrix;
  60         NameAssignment* nameMap;
  61         string filename, fastafile, listfile, namefile, groupfile, label, sorted, phylipfile, columnfile, distFile, format, outputDir, groups;
  62         ofstream out;
  63         ifstream in, inNames, inRow;
  64         bool abort, allLines, groupError, large, weighted;
  65         set<string> labels; //holds labels to be used
  66         map<string, int> nameToIndex;  //maps sequence name to index in sparsematrix
  67         map<string, string> nameFileMap;
  68         vector<string> outputNames, Groups;
  69         map<string, string> outputNameFiles;
  70         float cutoff;
  71         int precision;
  72         vector<SeqMap> seqVec;                  // contains maps with sequence index and distance
  73                                                                         // for all distances related to a certain sequence
  74         vector<int> rowPositions;
  75         map<string, vector<string> > outputTypes;
  76
  77         void readNamesFile();
  78         void readNamesFile(bool);
  79         int process(ListVector*);
  80         SeqMap getMap(int);
  81         string findRep(vector<string>);         // returns the name of the "representative" sequence of given bin or subset of a bin, for groups
  82         int processNames(string, string);
  83
  84
  85 };
  86
  87 #endif
  88