readdistcommand.cpp

   1 /*
   2  *  readdistcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/20/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "readdistcommand.h"
  11 #include "readphylip.h"
  12 #include "readcolumn.h"
  13 #include "readmatrix.hpp"
  14
  15 ReadDistCommand::ReadDistCommand(string option){
  16         try {
  17                 globaldata = GlobalData::getInstance();
  18                 abort = false;
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string Array[] =  {"phylip", "column", "name", "cutoff", "precision", "group"};
  26                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string, string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32
  33                         //check to make sure all parameters are valid for command
  34                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         globaldata->newRead();
  39
  40                         //check for required parameters
  41                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  42                         if (phylipfile == "not open") { abort = true; }
  43                         else if (phylipfile == "not found") { phylipfile = ""; }
  44                         else {  globaldata->setPhylipFile(phylipfile);  globaldata->setFormat("phylip");        }
  45
  46                         columnfile = validParameter.validFile(parameters, "column", true);
  47                         if (columnfile == "not open") { abort = true; }
  48                         else if (columnfile == "not found") { columnfile = ""; }
  49                         else {  globaldata->setColumnFile(columnfile); globaldata->setFormat("column"); }
  50
  51                         groupfile = validParameter.validFile(parameters, "group", true);
  52                         if (groupfile == "not open") { abort = true; }
  53                         else if (groupfile == "not found") { groupfile = ""; }
  54                         else {
  55                                 globaldata->setGroupFile(groupfile);
  56                                 //groupMap = new GroupMap(groupfile);
  57                                 //groupMap->readMap();
  58                         }
  59
  60                         namefile = validParameter.validFile(parameters, "name", true);
  61                         if (namefile == "not open") { abort = true; }
  62                         else if (namefile == "not found") { namefile = ""; }
  63                         else {  globaldata->setNameFile(namefile);      }
  64
  65                         //you are doing a list and group shared
  66                         if ((phylipfile != "") && (groupfile != "")) {
  67                         globaldata->setFormat("matrix"); }
  68
  69                         if ((phylipfile == "") && (columnfile == "")) { mothurOut("When executing a read.dist command you must enter a phylip or a column."); mothurOutEndLine(); abort = true; }
  70                         else if ((phylipfile != "") && (columnfile != "")) { mothurOut("When executing a read.dist command you must enter ONLY ONE of the following: phylip or column."); mothurOutEndLine(); abort = true; }
  71
  72                         if (columnfile != "") {
  73                                 if (namefile == "") {  cout << "You need to provide a namefile if you are going to use the column format." << endl; abort = true; }
  74                         }
  75
  76                         //check for optional parameter and set defaults
  77                         // ...at some point should added some additional type checking...
  78                         //get user cutoff and precision or use defaults
  79                         string temp;
  80                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
  81                         convert(temp, precision);
  82
  83                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
  84                         convert(temp, cutoff);
  85                         cutoff += (5 / (precision * 10.0));
  86
  87                         if (abort == false) {
  88                                 distFileName = globaldata->inputFileName;
  89                                 format = globaldata->getFormat();
  90
  91                                 if (format == "column") { read = new ReadColumnMatrix(distFileName); }
  92                                 else if (format == "phylip") { read = new ReadPhylipMatrix(distFileName); }
  93                                 else if (format == "matrix") {
  94                                         groupMap = new GroupMap(groupfile);
  95                                         groupMap->readMap();
  96
  97                                         if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap;  }
  98                                         globaldata->gGroupmap = groupMap;
  99                                 }
 100
 101                                 if (format != "matrix" ) {
 102                                         read->setCutoff(cutoff);
 103
 104                                         if(namefile != ""){
 105                                                 nameMap = new NameAssignment(namefile);
 106                                                 nameMap->readMap();
 107                                         }else{
 108                                                 nameMap = NULL;
 109                                         }
 110                                 }
 111                         }
 112
 113                 }
 114
 115         }
 116         catch(exception& e) {
 117                 errorOut(e, "ReadDistCommand", "ReadDistCommand");
 118                 exit(1);
 119         }
 120 }
 121 //**********************************************************************************************************************
 122
 123 void ReadDistCommand::help(){
 124         try {
 125                 mothurOut("The read.dist command parameter options are phylip or column, group, name, cutoff and precision\n");
 126                 mothurOut("The read.dist command can be used in two ways.  The first is to read a phylip or column and run the cluster command\n");
 127                 mothurOut("For this use the read.dist command should be in the following format: \n");
 128                 mothurOut("read.dist(phylip=yourDistFile, name=yourNameFile, cutoff=yourCutoff, precision=yourPrecision) \n");
 129                 mothurOut("The phylip or column parameter is required, but only one may be used.  If you use a column file the name filename is required. \n");
 130                 mothurOut("If you do not provide a cutoff value 10.00 is assumed. If you do not provide a precision value then 100 is assumed.\n");
 131                 mothurOut("The second way to use the read.dist command is to read a phylip or column and a group, so you can use the libshuff command.\n");
 132                 mothurOut("For this use the read.dist command should be in the following format: \n");
 133                 mothurOut("read.dist(phylip=yourPhylipfile, group=yourGroupFile). The cutoff and precision parameters are not valid with this use.  \n");
 134                 mothurOut("Note: No spaces between parameter labels (i.e. phylip), '=' and parameters (i.e.yourPhylipfile).\n\n");
 135         }
 136         catch(exception& e) {
 137                 errorOut(e, "ReadDistCommand", "help");
 138                 exit(1);
 139         }
 140 }
 141
 142 //**********************************************************************************************************************
 143
 144 ReadDistCommand::~ReadDistCommand(){
 145         if (abort == false) {
 146                 if (format != "matrix") {
 147                         delete read;
 148                         delete nameMap;
 149                 }
 150         }
 151 }
 152
 153 //**********************************************************************************************************************
 154 int ReadDistCommand::execute(){
 155         try {
 156
 157                 if (abort == true) {    return 0;       }
 158
 159                 time_t start = time(NULL);
 160                 size_t numDists = 0;
 161
 162                 if (format == "matrix") {
 163                         ifstream in;
 164                         openInputFile(distFileName, in);
 165                         matrix = new FullMatrix(in); //reads the matrix file
 166                         in.close();
 167                         //memory leak prevention
 168                         if (globaldata->gMatrix != NULL) { delete globaldata->gMatrix;  }
 169                         globaldata->gMatrix = matrix; //save matrix for coverage commands
 170                         numDists = matrix->getSizes()[1];
 171                 } else {
 172                         read->read(nameMap);
 173                         //to prevent memory leak
 174
 175                         if (globaldata->gListVector != NULL) {  delete globaldata->gListVector;  }
 176                         globaldata->gListVector = read->getListVector();
 177
 178                         if (globaldata->gSparseMatrix != NULL) { delete globaldata->gSparseMatrix;  }
 179                         globaldata->gSparseMatrix = read->getMatrix();
 180                         numDists = globaldata->gSparseMatrix->getNNodes();
 181
 182       int lines = cutoff / (1.0/precision);
 183       vector<float> dist_cutoff(lines+1,0);
 184                         for (int i = 0; i <= lines;i++) {
 185         dist_cutoff[i] = (i + 0.5) / precision;
 186       }
 187       vector<int> dist_count(lines+1,0);
 188       list<PCell>::iterator currentCell;
 189       SparseMatrix* smatrix = globaldata->gSparseMatrix;
 190                 for (currentCell = smatrix->begin(); currentCell != smatrix->end(); currentCell++) {
 191                                 for (int i = 0; i <= lines;i++) {
 192                                         if (currentCell->dist < dist_cutoff[i]) {
 193                                                 dist_count[i]++;
 194             break;
 195           }
 196         }
 197                         }
 198
 199      // string dist_string = "Dist:";
 200     //  string count_string = "Count: ";
 201                         //for (int i = 0; i <= lines;i++) {
 202         //dist_string = dist_string.append("\t").append(toString(dist_cutoff[i]));
 203       //        count_string = count_string.append("\t").append(toString(dist_count[i]));
 204                 //      }
 205       //mothurOut(dist_string); mothurOutEndLine(); mothurOut(count_string); mothurOutEndLine();
 206                 }
 207                 mothurOut("It took " + toString(time(NULL) - start) + " secs to read "); mothurOutEndLine();
 208                 return 0;
 209
 210         }
 211         catch(exception& e) {
 212                 errorOut(e, "ReadDistCommand", "execute");
 213                 exit(1);
 214         }
 215 }