countgroupscommand.cpp

   1 /*
   2  *  countgroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 8/9/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "countgroupscommand.h"
  11 #include "sharedutilities.h"
  12 #include "inputdata.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> CountGroupsCommand::setParameters(){
  16         try {
  17                 CommandParameter pshared("shared", "InputTypes", "", "", "sharedGroup", "sharedGroup", "none","summary",false,false,true); parameters.push_back(pshared);
  18                 CommandParameter pgroup("group", "InputTypes", "", "", "sharedGroup", "sharedGroup", "none","summary",false,false,true); parameters.push_back(pgroup);
  19         CommandParameter pcount("count", "InputTypes", "", "", "sharedGroup", "sharedGroup", "none","summary",false,false,true); parameters.push_back(pcount);
  20                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(paccnos);
  21                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "CountGroupsCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string CountGroupsCommand::getOutputPattern(string type) {
  36     try {
  37         string pattern = "";
  38
  39         if (type == "summary") {  pattern = "[filename],count.summary"; }
  40         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  41
  42         return pattern;
  43     }
  44     catch(exception& e) {
  45         m->errorOut(e, "PrimerDesignCommand", "getOutputPattern");
  46         exit(1);
  47     }
  48 }
  49 //**********************************************************************************************************************
  50 string CountGroupsCommand::getHelpString(){
  51         try {
  52                 string helpString = "";
  53                 helpString += "The count.groups command counts sequences from a specific group or set of groups from the following file types: group, count or shared file.\n";
  54                 helpString += "The count.groups command parameters are accnos, group, shared and groups. You must provide a group or shared file.\n";
  55                 helpString += "The accnos parameter allows you to provide a file containing the list of groups.\n";
  56                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like.  You can separate group names with dashes.\n";
  57                 helpString += "The count.groups command should be in the following format: count.groups(accnos=yourAccnos, group=yourGroupFile).\n";
  58                 helpString += "Example count.groups(accnos=amazon.accnos, group=amazon.groups).\n";
  59                 helpString += "or count.groups(groups=pasture, group=amazon.groups).\n";
  60                 helpString += "Note: No spaces between parameter labels (i.e. group), '=' and parameters (i.e.yourGroupFile).\n";
  61                 return helpString;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "CountGroupsCommand", "getHelpString");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 CountGroupsCommand::CountGroupsCommand(){
  70         try {
  71                 abort = true; calledHelp = true;
  72                 setParameters();
  73         vector<string> tempOutNames;
  74                 outputTypes["summary"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "CountGroupsCommand", "CountGroupsCommand");
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82 CountGroupsCommand::CountGroupsCommand(string option)  {
  83         try {
  84                 abort = false; calledHelp = false;
  85
  86                 //allow user to run help
  87                 if(option == "help") { help(); abort = true; calledHelp = true; }
  88                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  89
  90                 else {
  91                         vector<string> myArray = setParameters();
  92
  93                         OptionParser parser(option);
  94                         map<string,string> parameters = parser.getParameters();
  95
  96                         ValidParameters validParameter;
  97                         map<string,string>::iterator it;
  98
  99                         //check to make sure all parameters are valid for command
 100                         for (it = parameters.begin(); it != parameters.end(); it++) {
 101                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 102                         }
 103
 104                         //if the user changes the output directory command factory will send this info to us in the output parameter
 105                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 string path;
 112                                 it = parameters.find("accnos");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 118                                 }
 119
 120                                 it = parameters.find("group");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 126                                 }
 127
 128                                 it = parameters.find("shared");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 134                                 }
 135
 136                 it = parameters.find("count");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 142                                 }
 143                         }
 144
 145             vector<string> tempOutNames;
 146             outputTypes["summary"] = tempOutNames;
 147
 148                         //check for required parameters
 149                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 150                         if (accnosfile == "not open") { abort = true; }
 151                         else if (accnosfile == "not found") {  accnosfile = ""; }
 152                         else { m->setAccnosFile(accnosfile); }
 153
 154                         groups = validParameter.validFile(parameters, "groups", false);
 155                         if (groups == "not found") { groups = ""; }
 156                         else {
 157                                 m->splitAtDash(groups, Groups);
 158                                 m->setGroups(Groups);
 159                         }
 160
 161                         sharedfile = validParameter.validFile(parameters, "shared", true);
 162                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 163                         else if (sharedfile == "not found") {  sharedfile = "";  }
 164                         else { m->setSharedFile(sharedfile); }
 165
 166                         groupfile = validParameter.validFile(parameters, "group", true);
 167                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 168                         else if (groupfile == "not found") {    groupfile = ""; }
 169                         else { m->setGroupFile(groupfile); }
 170
 171             countfile = validParameter.validFile(parameters, "count", true);
 172             if (countfile == "not open") { countfile = ""; abort = true; }
 173             else if (countfile == "not found") { countfile = "";  }
 174             else {
 175                 m->setCountTableFile(countfile);
 176                 CountTable ct;
 177                 if (!ct.testGroups(countfile)) { m->mothurOut("[ERROR]: Your count file does not have any group information, aborting."); m->mothurOutEndLine(); abort=true; }
 178             }
 179
 180             if ((groupfile != "") && (countfile != "")) {
 181                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 182             }
 183
 184
 185                         if ((sharedfile == "") && (groupfile == "") && (countfile == "")) {
 186                                 //give priority to shared, then group
 187                                 sharedfile = m->getSharedFile();
 188                                 if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 189                                 else {
 190                                         groupfile = m->getGroupFile();
 191                                         if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 192                                         else {
 193                                                 countfile = m->getCountTableFile();
 194                         if (countfile != "") { m->mothurOut("Using " + countfile + " as input file for the count parameter."); m->mothurOutEndLine(); }
 195                         else {
 196                             m->mothurOut("You have no current groupfile, countfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 197                         }
 198                                         }
 199                                 }
 200                         }
 201
 202                         if ((accnosfile == "") && (Groups.size() == 0)) { Groups.push_back("all"); m->setGroups(Groups); }
 203                 }
 204
 205         }
 206         catch(exception& e) {
 207                 m->errorOut(e, "CountGroupsCommand", "CountGroupsCommand");
 208                 exit(1);
 209         }
 210 }
 211 //**********************************************************************************************************************
 212
 213 int CountGroupsCommand::execute(){
 214         try {
 215
 216                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 217
 218                 //get groups you want to remove
 219                 if (accnosfile != "") { m->readAccnos(accnosfile, Groups); m->setGroups(Groups); }
 220
 221                 if (groupfile != "") {
 222             map<string, string> variables;
 223             string thisOutputDir = outputDir;
 224             if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 225             variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(groupfile));
 226             string outputFileName = getOutputFileName("summary", variables);
 227             outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 228             ofstream out;
 229             m->openOutputFile(outputFileName, out);
 230
 231                         GroupMap groupMap(groupfile);
 232                         groupMap.readMap();
 233
 234                         //make sure groups are valid
 235                         //takes care of user setting groupNames that are invalid or setting groups=all
 236                         SharedUtil util;
 237                         vector<string> nameGroups = groupMap.getNamesOfGroups();
 238                         util.setGroups(Groups, nameGroups);
 239
 240             int total = 0;
 241                         for (int i = 0; i < Groups.size(); i++) {
 242                 int num = groupMap.getNumSeqs(Groups[i]);
 243                 total += num;
 244                                 m->mothurOut(Groups[i] + " contains " + toString(num) + "."); m->mothurOutEndLine();
 245                 out << Groups[i] << '\t' << num << endl;
 246                         }
 247             out.close();
 248             m->mothurOut("\nTotal seqs: " + toString(total) + "."); m->mothurOutEndLine();
 249                 }
 250
 251         if (m->control_pressed) { return 0; }
 252
 253         if (countfile != "") {
 254             map<string, string> variables;
 255             string thisOutputDir = outputDir;
 256             if (outputDir == "") {  thisOutputDir += m->hasPath(countfile);  }
 257             variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
 258             string outputFileName = getOutputFileName("summary", variables);
 259             outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 260             ofstream out;
 261             m->openOutputFile(outputFileName, out);
 262
 263                         CountTable ct;
 264                         ct.readTable(countfile, true);
 265
 266                         //make sure groups are valid
 267                         //takes care of user setting groupNames that are invalid or setting groups=all
 268                         SharedUtil util;
 269                         vector<string> nameGroups = ct.getNamesOfGroups();
 270                         util.setGroups(Groups, nameGroups);
 271
 272             int total = 0;
 273                         for (int i = 0; i < Groups.size(); i++) {
 274                 int num = ct.getGroupCount(Groups[i]);
 275                 total += num;
 276                                 m->mothurOut(Groups[i] + " contains " + toString(num) + "."); m->mothurOutEndLine();
 277                 out << Groups[i] << '\t' << num << endl;
 278                         }
 279             out.close();
 280
 281             m->mothurOut("\nTotal seqs: " + toString(total) + "."); m->mothurOutEndLine();
 282                 }
 283
 284                 if (m->control_pressed) { return 0; }
 285
 286                 if (sharedfile != "")           {
 287                         InputData input(sharedfile, "sharedfile");
 288                         vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 289
 290             map<string, string> variables;
 291             string thisOutputDir = outputDir;
 292             if (outputDir == "") {  thisOutputDir += m->hasPath(countfile);  }
 293             variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
 294             string outputFileName = getOutputFileName("summary", variables);
 295             outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 296             ofstream out;
 297             m->openOutputFile(outputFileName, out);
 298
 299             int total = 0;
 300                         for (int i = 0; i < lookup.size(); i++) {
 301                 int num = lookup[i]->getNumSeqs();
 302                 total += num;
 303                                 m->mothurOut(lookup[i]->getGroup() + " contains " + toString(num) + "."); m->mothurOutEndLine();
 304                                 delete lookup[i];
 305                 out << lookup[i]->getGroup() << '\t' << num << endl;
 306                         }
 307             out.close();
 308
 309             m->mothurOut("\nTotal seqs: " + toString(total) + "."); m->mothurOutEndLine();
 310                 }
 311
 312         m->mothurOutEndLine();
 313                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 314                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 315                 m->mothurOutEndLine();
 316
 317                 return 0;
 318         }
 319
 320         catch(exception& e) {
 321                 m->errorOut(e, "CountGroupsCommand", "execute");
 322                 exit(1);
 323         }
 324 }
 325 //**********************************************************************************************************************
 326
 327