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
  15 ChopSeqsCommand::ChopSeqsCommand(string option)  {
  16         try {
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"fasta","numbases","countgaps","keep","outputdir","inputdir"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string,string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31                         map<string,string>::iterator it;
  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                         //if the user changes the output directory command factory will send this info to us in the output parameter
  39                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  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("fasta");
  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["fasta"] = inputDir + it->second;            }
  52                                 }
  53                         }
  54
  55                         //check for required parameters
  56                         fastafile = validParameter.validFile(parameters, "fasta", true);
  57                         if (fastafile == "not open") { abort = true; }
  58                         else if (fastafile == "not found") {  m->mothurOut("You must provide a fasta file."); m->mothurOutEndLine(); abort = true; }
  59
  60                         string temp = validParameter.validFile(parameters, "numbases", false);  if (temp == "not found") { temp = "0"; }
  61                         convert(temp, numbases);
  62
  63                         temp = validParameter.validFile(parameters, "countgaps", false);        if (temp == "not found") { temp = "f"; }
  64                         countGaps = isTrue(temp);
  65
  66                         keep = validParameter.validFile(parameters, "keep", false);             if (keep == "not found") { keep = "front"; }
  67
  68                         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;  }
  69                 }
  70
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "ChopSeqsCommand", "ChopSeqsCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78
  79 void ChopSeqsCommand::help(){
  80         try {
  81                 m->mothurOut("The chop.seqs command reads a fasta file and outputs a .chop.fasta containing the trimmed sequences.\n");
  82                 m->mothurOut("The chop.seqs command parameters are fasta, numbases, countgaps and keep. fasta and numbases are required required.\n");
  83                 m->mothurOut("The chop.seqs command should be in the following format: chop.seqs(fasta=yourFasta, numbases=yourNum, keep=yourKeep).\n");
  84                 m->mothurOut("The numbases parameter allows you to specify the number of bases you want to keep.\n");
  85                 m->mothurOut("The keep parameter allows you to specify whether you want to keep the front or the back of your sequence, default=front.\n");
  86                 m->mothurOut("The countgaps parameter allows you to specify whether you want to count gaps as bases, default=false.\n");
  87                 m->mothurOut("Example chop.seqs(fasta=amazon.fasta, numbases=200, keep=front).\n");
  88                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  89         }
  90         catch(exception& e) {
  91                 m->errorOut(e, "ChopSeqsCommand", "help");
  92                 exit(1);
  93         }
  94 }
  95
  96 //**********************************************************************************************************************
  97
  98 int ChopSeqsCommand::execute(){
  99         try {
 100
 101                 if (abort == true) { return 0; }
 102
 103                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "chop.fasta";
 104                 string outputFileNameAccnos = outputDir + getRootName(getSimpleName(fastafile)) + "chop.accnos";
 105
 106                 ofstream out;
 107                 openOutputFile(outputFileName, out);
 108
 109                 ofstream outAcc;
 110                 openOutputFile(outputFileNameAccnos, outAcc);
 111
 112                 ifstream in;
 113                 openInputFile(fastafile, in);
 114
 115                 bool wroteAccnos = false;
 116
 117                 while (!in.eof()) {
 118
 119                         Sequence seq(in);
 120
 121                         if (m->control_pressed) { in.close(); out.close(); outAcc.close(); remove(outputFileName.c_str()); remove(outputFileNameAccnos.c_str()); return 0;  }
 122
 123                         if (seq.getName() != "") {
 124                                 string newSeqString = getChopped(seq);
 125
 126                                 //output trimmed sequence
 127                                 if (newSeqString != "") {
 128                                         out << ">" << seq.getName() << endl << newSeqString << endl;
 129                                 }else{
 130                                         outAcc << seq.getName() << endl;
 131                                         wroteAccnos = true;
 132                                 }
 133                         }
 134                 }
 135                 in.close();
 136                 out.close();
 137                 outAcc.close();
 138
 139                 m->mothurOutEndLine();
 140                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 141                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 142
 143                 if (wroteAccnos) { m->mothurOut(outputFileNameAccnos); m->mothurOutEndLine();  }
 144                 else {  remove(outputFileNameAccnos.c_str());  }
 145
 146                 m->mothurOutEndLine();
 147
 148                 return 0;
 149         }
 150
 151         catch(exception& e) {
 152                 m->errorOut(e, "ChopSeqsCommand", "execute");
 153                 exit(1);
 154         }
 155 }
 156 //**********************************************************************************************************************
 157 string ChopSeqsCommand::getChopped(Sequence seq) {
 158         try {
 159                 string temp = seq.getAligned();
 160                 string tempUnaligned = seq.getUnaligned();
 161
 162                 if (countGaps) {
 163                         //if needed trim sequence
 164                         if (keep == "front") {//you want to keep the beginning
 165                                 int tempLength = temp.length();
 166
 167                                 if (tempLength > numbases) { //you have enough bases to remove some
 168
 169                                         int stopSpot = 0;
 170                                         int numBasesCounted = 0;
 171
 172                                         for (int i = 0; i < temp.length(); i++) {
 173                                                 //eliminate N's
 174                                                 if (toupper(temp[i]) == 'N') { temp[i] == '.'; }
 175
 176                                                 numBasesCounted++;
 177
 178                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 179                                         }
 180
 181                                         if (stopSpot == 0) { temp = ""; }
 182                                         else {  temp = temp.substr(0, stopSpot);  }
 183
 184                                 }else { temp = ""; } //sequence too short
 185
 186                         }else { //you are keeping the back
 187                                 int tempLength = temp.length();
 188                                 if (tempLength > numbases) { //you have enough bases to remove some
 189
 190                                         int stopSpot = 0;
 191                                         int numBasesCounted = 0;
 192
 193                                         for (int i = (temp.length()-1); i >= 0; i--) {
 194                                                 //eliminate N's
 195                                                 if (toupper(temp[i]) == 'N') { temp[i] == '.'; }
 196
 197                                                 numBasesCounted++;
 198
 199                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 200                                         }
 201
 202                                         if (stopSpot == 0) { temp = ""; }
 203                                         else {  temp = temp.substr(stopSpot+1);  }
 204                                 }
 205                                 else {  temp = ""; } //sequence too short
 206                         }
 207
 208                 }else{
 209
 210                         //if needed trim sequence
 211                         if (keep == "front") {//you want to keep the beginning
 212                                 int tempLength = tempUnaligned.length();
 213
 214                                 if (tempLength > numbases) { //you have enough bases to remove some
 215
 216                                         int stopSpot = 0;
 217                                         int numBasesCounted = 0;
 218
 219                                         for (int i = 0; i < temp.length(); i++) {
 220                                                 //eliminate N's
 221                                                 if (toupper(temp[i]) == 'N') {
 222                                                         temp[i] == '.';
 223                                                         tempLength--;
 224                                                         if (tempLength < numbases) { stopSpot = 0; break; }
 225                                                 }
 226
 227                                                 if(isalpha(temp[i])) { numBasesCounted++; }
 228
 229                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 230                                         }
 231
 232                                         if (stopSpot == 0) { temp = ""; }
 233                                         else {  temp = temp.substr(0, stopSpot);  }
 234
 235                                 }else { temp = ""; } //sequence too short
 236
 237                         }else { //you are keeping the back
 238                                 int tempLength = tempUnaligned.length();
 239                                 if (tempLength > numbases) { //you have enough bases to remove some
 240
 241                                         int stopSpot = 0;
 242                                         int numBasesCounted = 0;
 243
 244                                         for (int i = (temp.length()-1); i >= 0; i--) {
 245                                                 //eliminate N's
 246                                                 if (toupper(temp[i]) == 'N') {
 247                                                         temp[i] == '.';
 248                                                         tempLength--;
 249                                                         if (tempLength < numbases) { stopSpot = 0; break; }
 250                                                 }
 251
 252                                                 if(isalpha(temp[i])) { numBasesCounted++; }
 253
 254                                                 if (numBasesCounted >= numbases) { stopSpot = i; break; }
 255                                         }
 256
 257                                         if (stopSpot == 0) { temp = ""; }
 258                                         else {  temp = temp.substr(stopSpot+1);  }
 259                                 }
 260                                 else {  temp = ""; } //sequence too short
 261                         }
 262                 }
 263
 264                 return temp;
 265         }
 266         catch(exception& e) {
 267                 m->errorOut(e, "ChopSeqsCommand", "getChopped");
 268                 exit(1);
 269         }
 270 }
 271 //**********************************************************************************************************************
 272
 273