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","outputdir","inputdir"};
  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                         map<string,string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         globaldata->newRead();
  40
  41                         //if the user changes the input directory command factory will send this info to us in the output parameter
  42                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  43                         if (inputDir == "not found"){   inputDir = "";          }
  44                         else {
  45                                 string path;
  46                                 it = parameters.find("phylip");
  47                                 //user has given a template file
  48                                 if(it != parameters.end()){
  49                                         path = hasPath(it->second);
  50                                         //if the user has not given a path then, add inputdir. else leave path alone.
  51                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
  52                                 }
  53
  54                                 it = parameters.find("column");
  55                                 //user has given a template file
  56                                 if(it != parameters.end()){
  57                                         path = hasPath(it->second);
  58                                         //if the user has not given a path then, add inputdir. else leave path alone.
  59                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
  60                                 }
  61
  62                                 it = parameters.find("name");
  63                                 //user has given a template file
  64                                 if(it != parameters.end()){
  65                                         path = hasPath(it->second);
  66                                         //if the user has not given a path then, add inputdir. else leave path alone.
  67                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  68                                 }
  69
  70                                 it = parameters.find("group");
  71                                 //user has given a template file
  72                                 if(it != parameters.end()){
  73                                         path = hasPath(it->second);
  74                                         //if the user has not given a path then, add inputdir. else leave path alone.
  75                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  76                                 }
  77                         }
  78
  79                         //if the user changes the output directory command factory will send this info to us in the output parameter
  80                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
  81
  82                         //check for required parameters
  83                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  84                         if (phylipfile == "not open") { abort = true; }
  85                         else if (phylipfile == "not found") { phylipfile = ""; }
  86                         else {  globaldata->setPhylipFile(phylipfile);  globaldata->setFormat("phylip");        }
  87
  88                         columnfile = validParameter.validFile(parameters, "column", true);
  89                         if (columnfile == "not open") { abort = true; }
  90                         else if (columnfile == "not found") { columnfile = ""; }
  91                         else {  globaldata->setColumnFile(columnfile); globaldata->setFormat("column"); }
  92
  93                         groupfile = validParameter.validFile(parameters, "group", true);
  94                         if (groupfile == "not open") { abort = true; }
  95                         else if (groupfile == "not found") { groupfile = ""; }
  96                         else {
  97                                 globaldata->setGroupFile(groupfile);
  98                                 //groupMap = new GroupMap(groupfile);
  99                                 //groupMap->readMap();
 100                         }
 101
 102                         namefile = validParameter.validFile(parameters, "name", true);
 103                         if (namefile == "not open") { abort = true; }
 104                         else if (namefile == "not found") { namefile = ""; }
 105                         else {  globaldata->setNameFile(namefile);      }
 106
 107                         //you are doing a list and group shared
 108                         if ((phylipfile != "") && (groupfile != "")) {
 109                         globaldata->setFormat("matrix"); }
 110
 111                         if ((phylipfile == "") && (columnfile == "")) { m->mothurOut("When executing a read.dist command you must enter a phylip or a column."); m->mothurOutEndLine(); abort = true; }
 112                         else if ((phylipfile != "") && (columnfile != "")) { m->mothurOut("When executing a read.dist command you must enter ONLY ONE of the following: phylip or column."); m->mothurOutEndLine(); abort = true; }
 113
 114                         if (columnfile != "") {
 115                                 if (namefile == "") {  cout << "You need to provide a namefile if you are going to use the column format." << endl; abort = true; }
 116                         }
 117
 118                         //check for optional parameter and set defaults
 119                         // ...at some point should added some additional type checking...
 120                         //get user cutoff and precision or use defaults
 121                         string temp;
 122                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
 123                         convert(temp, precision);
 124
 125                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 126                         convert(temp, cutoff);
 127                         cutoff += (5 / (precision * 10.0));
 128
 129                         if (abort == false) {
 130                                 distFileName = globaldata->inputFileName;
 131                                 format = globaldata->getFormat();
 132
 133                                 if (format == "column") { read = new ReadColumnMatrix(distFileName); }
 134                                 else if (format == "phylip") { read = new ReadPhylipMatrix(distFileName); }
 135                                 else if (format == "matrix") {
 136                                         groupMap = new GroupMap(groupfile);
 137                                         int error = groupMap->readMap();
 138                                         if (error == 1) { delete groupMap; abort = true; }
 139                                         else {
 140                                                 if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap;  }
 141                                                 globaldata->gGroupmap = groupMap;
 142                                         }
 143                                 }
 144
 145                                 if (format != "matrix" ) {
 146                                         read->setCutoff(cutoff);
 147
 148                                         if(namefile != ""){
 149                                                 nameMap = new NameAssignment(namefile);
 150                                                 nameMap->readMap();
 151                                         }else{
 152                                                 nameMap = NULL;
 153                                         }
 154                                 }
 155                         }
 156
 157                 }
 158
 159         }
 160         catch(exception& e) {
 161                 m->errorOut(e, "ReadDistCommand", "ReadDistCommand");
 162                 exit(1);
 163         }
 164 }
 165 //**********************************************************************************************************************
 166
 167 void ReadDistCommand::help(){
 168         try {
 169                 m->mothurOut("The read.dist command parameter options are phylip or column, group, name, cutoff and precision\n");
 170                 m->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");
 171                 m->mothurOut("For this use the read.dist command should be in the following format: \n");
 172                 m->mothurOut("read.dist(phylip=yourDistFile, name=yourNameFile, cutoff=yourCutoff, precision=yourPrecision) \n");
 173                 m->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");
 174                 m->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");
 175                 m->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");
 176                 m->mothurOut("For this use the read.dist command should be in the following format: \n");
 177                 m->mothurOut("read.dist(phylip=yourPhylipfile, group=yourGroupFile). The cutoff and precision parameters are not valid with this use.  \n");
 178                 m->mothurOut("Note: No spaces between parameter labels (i.e. phylip), '=' and parameters (i.e.yourPhylipfile).\n\n");
 179         }
 180         catch(exception& e) {
 181                 m->errorOut(e, "ReadDistCommand", "help");
 182                 exit(1);
 183         }
 184 }
 185
 186 //**********************************************************************************************************************
 187
 188 ReadDistCommand::~ReadDistCommand(){
 189         if (abort == false) {
 190                 if (format != "matrix") {
 191                         delete read;
 192                         delete nameMap;
 193                 }
 194         }
 195 }
 196
 197 //**********************************************************************************************************************
 198 int ReadDistCommand::execute(){
 199         try {
 200
 201                 if (abort == true) {    return 0;       }
 202
 203                 time_t start = time(NULL);
 204                 size_t numDists = 0;
 205
 206                 vector<string> outputNames;
 207
 208                 if (format == "matrix") {
 209                         ifstream in;
 210                         openInputFile(distFileName, in);
 211                         matrix = new FullMatrix(in); //reads the matrix file
 212                         in.close();
 213
 214                         //if files don't match...
 215                         if (matrix->getNumSeqs() < groupMap->getNumSeqs()) {
 216                                 m->mothurOut("Your distance file contains " + toString(matrix->getNumSeqs()) + " sequences, and your group file contains " + toString(groupMap->getNumSeqs()) + " sequences.");  m->mothurOutEndLine();
 217                                 //create new group file
 218                                 if(outputDir == "") { outputDir += hasPath(groupfile); }
 219
 220                                 string newGroupFile = outputDir + getRootName(getSimpleName(groupfile)) + "editted.groups";
 221                                 outputNames.push_back(newGroupFile);
 222                                 ofstream outGroups;
 223                                 openOutputFile(newGroupFile, outGroups);
 224
 225                                 for (int i = 0; i < matrix->getNumSeqs(); i++) {
 226                                         Names temp = matrix->getRowInfo(i);
 227                                         outGroups << temp.seqName << '\t' << temp.groupName << endl;
 228                                 }
 229                                 outGroups.close();
 230
 231                                 m->mothurOut(newGroupFile + " is a new group file containing only the sequence that are in your distance file. I will read this file instead."); m->mothurOutEndLine();
 232
 233                                 //read new groupfile
 234                                 delete groupMap; groupMap = NULL;
 235                                 groupfile = newGroupFile;
 236                                 globaldata->setGroupFile(groupfile);
 237
 238                                 groupMap = new GroupMap(groupfile);
 239                                 groupMap->readMap();
 240
 241                                 globaldata->gGroupmap = groupMap;
 242                         }
 243
 244                         //memory leak prevention
 245                         if (globaldata->gMatrix != NULL) { delete globaldata->gMatrix;  }
 246                         globaldata->gMatrix = matrix; //save matrix for coverage commands
 247                         numDists = matrix->getSizes()[1];
 248                 } else {
 249                         read->read(nameMap);
 250                         //to prevent memory leak
 251
 252                         if (globaldata->gListVector != NULL) {  delete globaldata->gListVector;  }
 253                         globaldata->gListVector = read->getListVector();
 254
 255                         if (globaldata->gSparseMatrix != NULL) { delete globaldata->gSparseMatrix;  }
 256                         globaldata->gSparseMatrix = read->getMatrix();
 257                         numDists = globaldata->gSparseMatrix->getNNodes();
 258         //cout << "matrix contains " << numDists << " distances." << endl;
 259
 260     /*  int lines = cutoff / (1.0/precision);
 261       vector<float> dist_cutoff(lines+1,0);
 262                         for (int i = 0; i <= lines;i++) {
 263         dist_cutoff[i] = (i + 0.5) / precision;
 264       }
 265       vector<int> dist_count(lines+1,0);
 266       list<PCell>::iterator currentCell;
 267       SparseMatrix* smatrix = globaldata->gSparseMatrix;
 268                 for (currentCell = smatrix->begin(); currentCell != smatrix->end(); currentCell++) {
 269                                 for (int i = 0; i <= lines;i++) {
 270                                         if (currentCell->dist < dist_cutoff[i]) {
 271                                                 dist_count[i]++;
 272             break;
 273           }
 274         }
 275                         }
 276 */
 277      // string dist_string = "Dist:";
 278     //  string count_string = "Count: ";
 279                         //for (int i = 0; i <= lines;i++) {
 280         //dist_string = dist_string.append("\t").append(toString(dist_cutoff[i]));
 281       //        count_string = count_string.append("\t").append(toString(dist_count[i]));
 282                 //      }
 283       //m->mothurOut(dist_string); m->mothurOutEndLine(); m->mothurOut(count_string); m->mothurOutEndLine();
 284                 }
 285
 286                 if (outputNames.size() != 0) {
 287                         m->mothurOutEndLine();
 288                         m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 289                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 290                         m->mothurOutEndLine();
 291                 }
 292
 293                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to read "); m->mothurOutEndLine();
 294                 return 0;
 295
 296         }
 297         catch(exception& e) {
 298                 m->errorOut(e, "ReadDistCommand", "execute");
 299                 exit(1);
 300         }
 301 }