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 "listvector.hpp"
  16 #include "inputdata.h"
  17 #include "fastamap.h"
  18 #include "groupmap.h"
  19 #include "readmatrix.hpp"
  20 #include "formatmatrix.h"
  21 #include "counttable.h"
  22
  23 typedef map<int, float> SeqMap;
  24
  25 struct repStruct {
  26                 string name;
  27                 int     bin;
  28                 int size;
  29                 string group;
  30
  31                 repStruct(){}
  32                 repStruct(string n, int b, int s, string g) : name(n), bin(b), size(s), group(g) {}
  33                 ~repStruct() {}
  34 };
  35
  36 class GetOTURepCommand : public Command {
  37
  38 public:
  39         GetOTURepCommand(string);
  40         GetOTURepCommand();
  41         ~GetOTURepCommand(){}
  42
  43         vector<string> setParameters();
  44         string getCommandName()                 { return "get.oturep";                          }
  45         string getCommandCategory()             { return "OTU-Based Approaches";        }
  46
  47         string getHelpString();
  48     string getOutputPattern(string);
  49         string getCitation() { return "http://www.mothur.org/wiki/Get.oturep"; }
  50         string getDescription()         { return "gets a representative sequence for each OTU"; }
  51
  52
  53         int execute();
  54         void help() { m->mothurOut(getHelpString()); }
  55
  56
  57 private:
  58         ListVector* list;
  59         InputData* input;
  60         FastaMap* fasta;
  61         GroupMap* groupMap;
  62         ReadMatrix* readMatrix;
  63         FormatMatrix* formatMatrix;
  64         NameAssignment* nameMap;
  65     CountTable ct;
  66         string filename, fastafile, listfile, namefile, groupfile, label, sorted, phylipfile, countfile, columnfile, distFile, format, outputDir, groups;
  67         ofstream out;
  68         ifstream in, inNames, inRow;
  69         bool abort, allLines, groupError, large, weighted, hasGroups;
  70         set<string> labels; //holds labels to be used
  71         map<string, int> nameToIndex;  //maps sequence name to index in sparsematrix
  72         map<string, string> nameFileMap;
  73         vector<string> outputNames, Groups;
  74         map<string, string> outputNameFiles;
  75         float cutoff;
  76         int precision;
  77         vector<SeqMap> seqVec;                  // contains maps with sequence index and distance
  78                                                                         // for all distances related to a certain sequence
  79         vector<int> rowPositions;
  80
  81         void readNamesFile();
  82         void readNamesFile(bool);
  83         int process(ListVector*);
  84         SeqMap getMap(int);
  85         string findRep(vector<string>, string);         // returns the name of the "representative" sequence of given bin or subset of a bin, for groups
  86         int processNames(string, string);
  87         int processFastaNames(string, string);
  88     int readDist();
  89 };
  90
  91 #endif
  92