chopseqscommand.cpp

   1 /*
   2  *  chopseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 5/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chopseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ChopSeqsCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pnumbases("numbases", "Number", "", "0", "", "", "",false,true); parameters.push_back(pnumbases);
  18                 CommandParameter pcountgaps("countgaps", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pcountgaps);
  19                 CommandParameter pshort("short", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pshort);
  20                 CommandParameter pkeep("keep", "Multiple", "front-back", "front", "", "", "",false,false); parameters.push_back(pkeep);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "ChopSeqsCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string ChopSeqsCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The chop.seqs command reads a fasta file and outputs a .chop.fasta containing the trimmed sequences. Note: If a sequence is completely 'chopped', an accnos file will be created with the names of the sequences removed. \n";
  38                 helpString += "The chop.seqs command parameters are fasta, numbases, countgaps and keep. fasta is required unless you have a valid current fasta file. numbases is required.\n";
  39                 helpString += "The chop.seqs command should be in the following format: chop.seqs(fasta=yourFasta, numbases=yourNum, keep=yourKeep).\n";
  40                 helpString += "The numbases parameter allows you to specify the number of bases you want to keep.\n";
  41                 helpString += "The keep parameter allows you to specify whether you want to keep the front or the back of your sequence, default=front.\n";
  42                 helpString += "The countgaps parameter allows you to specify whether you want to count gaps as bases, default=false.\n";
  43                 helpString += "The short parameter allows you to specify you want to keep sequences that are too short to chop, default=false.\n";
  44                 helpString += "For example, if you ran chop.seqs with numbases=200 and short=t, if a sequence had 100 bases mothur would keep the sequence rather than eliminate it.\n";
  45                 helpString += "Example chop.seqs(fasta=amazon.fasta, numbases=200, keep=front).\n";
  46                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  47                 return helpString;
  48         }
  49         catch(exception& e) {
  50                 m->errorOut(e, "ChopSeqsCommand", "getHelpString");
  51                 exit(1);
  52         }
  53 }
  54
  55 //**********************************************************************************************************************
  56 ChopSeqsCommand::ChopSeqsCommand(){
  57         try {
  58                 abort = true; calledHelp = true;
  59                 setParameters();
  60                 vector<string> tempOutNames;
  61                 outputTypes["fasta"] = tempOutNames;
  62                 outputTypes["accnos"] = tempOutNames;
  63         }
  64         catch(exception& e) {
  65                 m->errorOut(e, "ChopSeqsCommand", "ChopSeqsCommand");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70 ChopSeqsCommand::ChopSeqsCommand(string option)  {
  71         try {
  72                 abort = false; calledHelp = false;
  73
  74                 //allow user to run help
  75                 if(option == "help") { help(); abort = true; calledHelp = true; }
  76                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  77
  78                 else {
  79                         vector<string> myArray = setParameters();
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83
  84                         ValidParameters validParameter;
  85                         map<string,string>::iterator it;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95                         outputTypes["accnos"] = tempOutNames;
  96
  97                         //if the user changes the input directory command factory will send this info to us in the output parameter
  98                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  99                         if (inputDir == "not found"){   inputDir = "";          }
 100                         else {
 101                                 string path;
 102                                 it = parameters.find("fasta");
 103                                 //user has given a template file
 104                                 if(it != parameters.end()){
 105                                         path = m->hasPath(it->second);
 106                                         //if the user has not given a path then, add inputdir. else leave path alone.
 107                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 108                                 }
 109                         }
 110
 111                         //check for required parameters
 112                         fastafile = validParameter.validFile(parameters, "fasta", true);
 113                         if (fastafile == "not open") { abort = true; }
 114                         else if (fastafile == "not found") {                            //if there is a current fasta file, use it
 115                                 fastafile = m->getFastaFile();
 116                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 117                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 118                         }else { m->setFastaFile(fastafile); }
 119
 120                         //if the user changes the output directory command factory will send this info to us in the output parameter
 121                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 122
 123                         string temp = validParameter.validFile(parameters, "numbases", false);  if (temp == "not found") { temp = "0"; }
 124                         m->mothurConvert(temp, numbases);
 125
 126                         temp = validParameter.validFile(parameters, "countgaps", false);        if (temp == "not found") { temp = "f"; }
 127                         countGaps = m->isTrue(temp);
 128
 129                         temp = validParameter.validFile(parameters, "short", false);    if (temp == "not found") { temp = "f"; }
 130                         Short = m->isTrue(temp);
 131
 132                         keep = validParameter.validFile(parameters, "keep", false);             if (keep == "not found") { keep = "front"; }
 133
 134                         if (numbases == 0)  { m->mothurOut("You must provide the number of bases you want to keep for the chops.seqs command."); m->mothurOutEndLine(); abort = true;  }
 135                 }
 136
 137         }
 138         catch(exception& e) {
 139                 m->errorOut(e, "ChopSeqsCommand", "ChopSeqsCommand");
 140                 exit(1);
 141         }
 142 }
 143 //**********************************************************************************************************************
 144
 145 int ChopSeqsCommand::execute(){
 146         try {
 147
 148                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 149
 150                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "chop.fasta";
 151                 string outputFileNameAccnos = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "chop.accnos";
 152
 153                 ofstream out;
 154                 m->openOutputFile(outputFileName, out);
 155
 156                 ofstream outAcc;
 157                 m->openOutputFile(outputFileNameAccnos, outAcc);
 158
 159                 ifstream in;
 160                 m->openInputFile(fastafile, in);
 161
 162                 bool wroteAccnos = false;
 163
 164                 while (!in.eof()) {
 165
 166                         Sequence seq(in);
 167
 168                         if (m->control_pressed) { outputTypes.clear(); in.close(); out.close(); outAcc.close(); m->mothurRemove(outputFileName); m->mothurRemove(outputFileNameAccnos); return 0;  }
 169
 170                         if (seq.getName() != "") {
 171                                 string newSeqString = getChopped(seq);
 172
 173                                 //output trimmed sequence
 174                                 if (newSeqString != "") {
 175                                         out << ">" << seq.getName() << endl << newSeqString << endl;
 176                                 }else{
 177                                         outAcc << seq.getName() << endl;
 178                                         wroteAccnos = true;
 179                                 }
 180                         }
 181                 }
 182                 in.close();
 183                 out.close();
 184                 outAcc.close();
 185
 186                 m->mothurOutEndLine();
 187                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 188                 m->mothurOut(outputFileName); m->mothurOutEndLine();    outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 189
 190                 if (wroteAccnos) { m->mothurOut(outputFileNameAccnos); m->mothurOutEndLine(); outputNames.push_back(outputFileNameAccnos); outputTypes["accnos"].push_back(outputFileNameAccnos); }
 191                 else {  m->mothurRemove(outputFileNameAccnos);  }
 192
 193                 m->mothurOutEndLine();
 194
 195                 //set fasta file as new current fastafile
 196                 string current = "";
 197                 itTypes = outputTypes.find("fasta");
 198                 if (itTypes != outputTypes.end()) {
 199                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 200                 }
 201
 202                 if (wroteAccnos) { //set accnos file as new current accnosfile
 203                         itTypes = outputTypes.find("accnos");
 204                         if (itTypes != outputTypes.end()) {
 205                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 206                         }
 207                 }
 208
 209
 210                 return 0;
 211         }
 212
 213         catch(exception& e) {
 214                 m->errorOut(e, "ChopSeqsCommand", "execute");
 215                 exit(1);
 216         }
 217 }
 218 //**********************************************************************************************************************
 219 string ChopSeqsCommand::getChopped(Sequence seq) {
 220         try {
 221                 string temp = seq.getAligned();
 222                 string tempUnaligned = seq.getUnaligned();
 223
 224                 if (countGaps) {
 225                         //if needed trim sequence
 226                         if (keep == "front") {//you want to keep the beginning
 227                                 int tempLength = temp.length();
 228
 229                                 if (tempLength > numbases) { //you have enough bases to remove some
 230
 231                                         int stopSpot = 0;
 232                                         int numBasesCounted = 0;
 233
 234                                         for (int i = 0; i < temp.length(); i++) {
 235                                                 //eliminate N's
 236                                                 if (toupper(temp[i]) == 'N') { temp[i] = '.'; }
 237
 238                                                 numBasesCounted++;
 239
 240                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 241                                         }
 242
 243                                         if (stopSpot == 0) { temp = ""; }
 244                                         else {  temp = temp.substr(0, stopSpot+1);  }
 245
 246                                 }else {
 247                                         if (!Short) { temp = ""; } //sequence too short
 248                                 }
 249                         }else { //you are keeping the back
 250                                 int tempLength = temp.length();
 251                                 if (tempLength > numbases) { //you have enough bases to remove some
 252
 253                                         int stopSpot = 0;
 254                                         int numBasesCounted = 0;
 255
 256                                         for (int i = (temp.length()-1); i >= 0; i--) {
 257                                                 //eliminate N's
 258                                                 if (toupper(temp[i]) == 'N') { temp[i] = '.'; }
 259
 260                                                 numBasesCounted++;
 261
 262                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 263                                         }
 264
 265                                         if (stopSpot == 0) { temp = ""; }
 266                                         else {  temp = temp.substr(stopSpot+1);  }
 267                                 }else {
 268                                         if (!Short) { temp = ""; } //sequence too short
 269                                 }
 270                         }
 271
 272                 }else{
 273
 274                         //if needed trim sequence
 275                         if (keep == "front") {//you want to keep the beginning
 276                                 int tempLength = tempUnaligned.length();
 277
 278                                 if (tempLength > numbases) { //you have enough bases to remove some
 279
 280                                         int stopSpot = 0;
 281                                         int numBasesCounted = 0;
 282
 283                                         for (int i = 0; i < temp.length(); i++) {
 284                                                 //eliminate N's
 285                                                 if (toupper(temp[i]) == 'N') {
 286                                                         temp[i] = '.';
 287                                                         tempLength--;
 288                                                         if (tempLength < numbases) { stopSpot = 0; break; }
 289                                                 }
 290
 291                                                 if(isalpha(temp[i])) { numBasesCounted++; }
 292
 293                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 294                                         }
 295
 296                                         if (stopSpot == 0) { temp = ""; }
 297                                         else {  temp = temp.substr(0, stopSpot+1);  }
 298
 299                                 }else {
 300                                         if (!Short) { temp = ""; } //sequence too short
 301                                 }
 302                         }else { //you are keeping the back
 303                                 int tempLength = tempUnaligned.length();
 304                                 if (tempLength > numbases) { //you have enough bases to remove some
 305
 306                                         int stopSpot = 0;
 307                                         int numBasesCounted = 0;
 308
 309                                         for (int i = (temp.length()-1); i >= 0; i--) {
 310                                                 //eliminate N's
 311                                                 if (toupper(temp[i]) == 'N') {
 312                                                         temp[i] = '.';
 313                                                         tempLength--;
 314                                                         if (tempLength < numbases) { stopSpot = 0; break; }
 315                                                 }
 316
 317                                                 if(isalpha(temp[i])) { numBasesCounted++; }
 318
 319                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 320                                         }
 321
 322                                         if (stopSpot == 0) { temp = ""; }
 323                                         else {  temp = temp.substr(stopSpot);  }
 324                                 }else {
 325                                         if (!Short) { temp = ""; } //sequence too short
 326                                 }
 327                         }
 328                 }
 329
 330                 return temp;
 331         }
 332         catch(exception& e) {
 333                 m->errorOut(e, "ChopSeqsCommand", "getChopped");
 334                 exit(1);
 335         }
 336 }
 337 //**********************************************************************************************************************
 338
 339