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","sim"};
  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, "sim", false);                              if (temp == "not found") { temp = "F"; }
 126                         sim = isTrue(temp);
 127                         globaldata->sim = sim;
 128
 129                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 130                         convert(temp, cutoff);
 131                         cutoff += (5 / (precision * 10.0));
 132
 133                         if (abort == false) {
 134                                 distFileName = globaldata->inputFileName;
 135                                 format = globaldata->getFormat();
 136
 137                                 if (format == "column") { read = new ReadColumnMatrix(distFileName); }
 138                                 else if (format == "phylip") { read = new ReadPhylipMatrix(distFileName); }
 139                                 else if (format == "matrix") {
 140                                         groupMap = new GroupMap(groupfile);
 141                                         int error = groupMap->readMap();
 142                                         if (error == 1) { delete groupMap; abort = true; }
 143                                         else {
 144                                                 if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap;  }
 145                                                 globaldata->gGroupmap = groupMap;
 146                                         }
 147                                 }
 148
 149                                 if (format != "matrix" ) {
 150                                         read->setCutoff(cutoff);
 151
 152                                         if(namefile != ""){
 153                                                 nameMap = new NameAssignment(namefile);
 154                                                 nameMap->readMap();
 155                                         }else{
 156                                                 nameMap = NULL;
 157                                         }
 158                                 }
 159                         }
 160
 161                 }
 162
 163         }
 164         catch(exception& e) {
 165                 m->errorOut(e, "ReadDistCommand", "ReadDistCommand");
 166                 exit(1);
 167         }
 168 }
 169 //**********************************************************************************************************************
 170
 171 void ReadDistCommand::help(){
 172         try {
 173                 m->mothurOut("The read.dist command parameter options are phylip or column, group, name, sim, cutoff and precision\n");
 174                 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");
 175                 m->mothurOut("For this use the read.dist command should be in the following format: \n");
 176                 m->mothurOut("read.dist(phylip=yourDistFile, name=yourNameFile, cutoff=yourCutoff, precision=yourPrecision) \n");
 177                 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");
 178                 m->mothurOut("The sim parameter is used to indicate that your distance file contains similarity values instead of distance values. The default is false, if sim=true then mothur will convert the similarity values to distances. \n");
 179                 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");
 180                 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");
 181                 m->mothurOut("For this use the read.dist command should be in the following format: \n");
 182                 m->mothurOut("read.dist(phylip=yourPhylipfile, group=yourGroupFile). The cutoff and precision parameters are not valid with this use.  \n");
 183                 m->mothurOut("Note: No spaces between parameter labels (i.e. phylip), '=' and parameters (i.e.yourPhylipfile).\n\n");
 184         }
 185         catch(exception& e) {
 186                 m->errorOut(e, "ReadDistCommand", "help");
 187                 exit(1);
 188         }
 189 }
 190
 191 //**********************************************************************************************************************
 192
 193 ReadDistCommand::~ReadDistCommand(){
 194         if (abort == false) {
 195                 if (format != "matrix") {
 196                         delete read;
 197                         delete nameMap;
 198                 }
 199         }
 200 }
 201
 202 //**********************************************************************************************************************
 203 int ReadDistCommand::execute(){
 204         try {
 205
 206                 if (abort == true) {    return 0;       }
 207
 208                 time_t start = time(NULL);
 209                 size_t numDists = 0;
 210
 211                 vector<string> outputNames;
 212
 213                 if (format == "matrix") {
 214                         ifstream in;
 215                         openInputFile(distFileName, in);
 216                         matrix = new FullMatrix(in); //reads the matrix file
 217                         in.close();
 218
 219                         if (m->control_pressed) { delete groupMap; delete matrix; return 0; }
 220
 221                         //if files don't match...
 222                         if (matrix->getNumSeqs() < groupMap->getNumSeqs()) {
 223                                 m->mothurOut("Your distance file contains " + toString(matrix->getNumSeqs()) + " sequences, and your group file contains " + toString(groupMap->getNumSeqs()) + " sequences.");  m->mothurOutEndLine();
 224                                 //create new group file
 225                                 if(outputDir == "") { outputDir += hasPath(groupfile); }
 226
 227                                 string newGroupFile = outputDir + getRootName(getSimpleName(groupfile)) + "editted.groups";
 228                                 outputNames.push_back(newGroupFile);
 229                                 ofstream outGroups;
 230                                 openOutputFile(newGroupFile, outGroups);
 231
 232                                 for (int i = 0; i < matrix->getNumSeqs(); i++) {
 233                                         if (m->control_pressed) { delete groupMap; delete matrix; outGroups.close(); remove(newGroupFile.c_str()); return 0; }
 234
 235                                         Names temp = matrix->getRowInfo(i);
 236                                         outGroups << temp.seqName << '\t' << temp.groupName << endl;
 237                                 }
 238                                 outGroups.close();
 239
 240                                 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();
 241
 242                                 //read new groupfile
 243                                 delete groupMap; groupMap = NULL;
 244                                 groupfile = newGroupFile;
 245                                 globaldata->setGroupFile(groupfile);
 246
 247                                 groupMap = new GroupMap(groupfile);
 248                                 groupMap->readMap();
 249
 250                                 if (m->control_pressed) { delete groupMap; delete matrix; remove(newGroupFile.c_str()); return 0; }
 251
 252                                 globaldata->gGroupmap = groupMap;
 253                         }
 254
 255                         //memory leak prevention
 256                         if (globaldata->gMatrix != NULL) { delete globaldata->gMatrix;  }
 257                         globaldata->gMatrix = matrix; //save matrix for coverage commands
 258                         numDists = matrix->getSizes()[1];
 259                 } else {
 260                         read->read(nameMap);
 261                         //to prevent memory leak
 262
 263                         if (m->control_pressed) {  return 0; }
 264
 265                         if (globaldata->gListVector != NULL) {  delete globaldata->gListVector;  }
 266                         globaldata->gListVector = read->getListVector();
 267
 268                         if (globaldata->gSparseMatrix != NULL) { delete globaldata->gSparseMatrix;  }
 269                         globaldata->gSparseMatrix = read->getMatrix();
 270                         numDists = globaldata->gSparseMatrix->getNNodes();
 271                 }
 272
 273                 if (m->control_pressed) {  return 0; }
 274
 275                 if (outputNames.size() != 0) {
 276                         m->mothurOutEndLine();
 277                         m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 278                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 279                         m->mothurOutEndLine();
 280                 }
 281
 282                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to read "); m->mothurOutEndLine();
 283                 return 0;
 284
 285         }
 286         catch(exception& e) {
 287                 m->errorOut(e, "ReadDistCommand", "execute");
 288                 exit(1);
 289         }
 290 }