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","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                         keep = validParameter.validFile(parameters, "keep", false);             if (keep == "not found") { keep = "front"; }
  64
  65                         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;  }
  66                 }
  67
  68         }
  69         catch(exception& e) {
  70                 m->errorOut(e, "ChopSeqsCommand", "ChopSeqsCommand");
  71                 exit(1);
  72         }
  73 }
  74 //**********************************************************************************************************************
  75
  76 void ChopSeqsCommand::help(){
  77         try {
  78                 m->mothurOut("The chop.seqs command reads a fasta file and outputs a .chop.fasta containing the trimmed sequences.\n");
  79                 m->mothurOut("The chop.seqs command parameters are fasta, numbases, and keep. fasta and numbases are required required.\n");
  80                 m->mothurOut("The chop.seqs command should be in the following format: chop.seqs(fasta=yourFasta, numbases=yourNum, keep=yourKeep).\n");
  81                 m->mothurOut("The numbases parameter allows you to specify the number of bases you want to keep.\n");
  82                 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");
  83                 m->mothurOut("Example chop.seqs(fasta=amazon.fasta, numbases=200, keep=front).\n");
  84                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "ChopSeqsCommand", "help");
  88                 exit(1);
  89         }
  90 }
  91
  92 //**********************************************************************************************************************
  93
  94 int ChopSeqsCommand::execute(){
  95         try {
  96
  97                 if (abort == true) { return 0; }
  98
  99                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "chop.fasta";
 100                 string outputFileNameAccnos = outputDir + getRootName(getSimpleName(fastafile)) + "chop.accnos";
 101
 102                 ofstream out;
 103                 openOutputFile(outputFileName, out);
 104
 105                 ofstream outAcc;
 106                 openOutputFile(outputFileNameAccnos, outAcc);
 107
 108                 ifstream in;
 109                 openInputFile(fastafile, in);
 110
 111                 bool wroteAccnos = false;
 112
 113                 while (!in.eof()) {
 114
 115                         Sequence seq(in);
 116
 117                         if (m->control_pressed) { in.close(); out.close(); outAcc.close(); remove(outputFileName.c_str()); remove(outputFileNameAccnos.c_str()); return 0;  }
 118
 119                         if (seq.getName() != "") {
 120                                 string newSeqString = getChopped(seq);
 121
 122                                 //output trimmed sequence
 123                                 if (newSeqString != "") {
 124                                         out << ">" << seq.getName() << endl << newSeqString << endl;
 125                                 }else{
 126                                         outAcc << seq.getName() << endl;
 127                                         wroteAccnos = true;
 128                                 }
 129                         }
 130                 }
 131                 in.close();
 132                 out.close();
 133                 outAcc.close();
 134
 135                 m->mothurOutEndLine();
 136                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 137                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 138
 139                 if (wroteAccnos) { m->mothurOut(outputFileNameAccnos); m->mothurOutEndLine();  }
 140                 else {  remove(outputFileNameAccnos.c_str());  }
 141
 142                 m->mothurOutEndLine();
 143
 144                 return 0;
 145         }
 146
 147         catch(exception& e) {
 148                 m->errorOut(e, "ChopSeqsCommand", "execute");
 149                 exit(1);
 150         }
 151 }
 152 //**********************************************************************************************************************
 153 string ChopSeqsCommand::getChopped(Sequence seq) {
 154         try {
 155                 string temp = seq.getAligned();
 156                 string tempUnaligned = seq.getUnaligned();
 157
 158                 //if needed trim sequence
 159                 if (keep == "front") {//you want to keep the beginning
 160                         int tempLength = tempUnaligned.length();
 161
 162                         if (tempLength > numbases) { //you have enough bases to remove some
 163
 164                                 int stopSpot = 0;
 165                                 int numBasesCounted = 0;
 166
 167                                 for (int i = 0; i < temp.length(); i++) {
 168                                         //eliminate N's
 169                                         if (toupper(temp[i]) == 'N') {
 170                                                 temp[i] == '.';
 171                                                 tempLength--;
 172                                                 if (tempLength < numbases) { stopSpot = 0; break; }
 173                                         }
 174
 175                                         if(isalpha(temp[i])) { numBasesCounted++; }
 176
 177                                         if (numBasesCounted >= numbases) { stopSpot = i; break; }
 178                                 }
 179
 180                                 if (stopSpot == 0) { temp = ""; }
 181                                 else {  temp = temp.substr(0, stopSpot);  }
 182
 183                         }else { temp = ""; } //sequence too short
 184
 185                 }else { //you are keeping the back
 186                         int tempLength = tempUnaligned.length();
 187                         if (tempLength > numbases) { //you have enough bases to remove some
 188
 189                                 int stopSpot = 0;
 190                                 int numBasesCounted = 0;
 191
 192                                 for (int i = (temp.length()-1); i >= 0; i--) {
 193                                         //eliminate N's
 194                                         if (toupper(temp[i]) == 'N') {
 195                                                 temp[i] == '.';
 196                                                 tempLength--;
 197                                                 if (tempLength < numbases) { stopSpot = 0; break; }
 198                                         }
 199
 200                                         if(isalpha(temp[i])) { numBasesCounted++; }
 201
 202                                         if (numBasesCounted >= numbases) { stopSpot = i; break; }
 203                                 }
 204
 205                                 if (stopSpot == 0) { temp = ""; }
 206                                 else {  temp = temp.substr(stopSpot+1);  }
 207                         }
 208                         else {  temp = ""; } //sequence too short
 209                 }
 210
 211                 return temp;
 212         }
 213         catch(exception& e) {
 214                 m->errorOut(e, "ChopSeqsCommand", "getChopped");
 215                 exit(1);
 216         }
 217 }
 218 //**********************************************************************************************************************
 219
 220