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","end","fromend","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, "end", false);       if (temp == "not found") { temp = "0"; }
  61                         convert(temp, end);
  62
  63                         temp = validParameter.validFile(parameters, "fromend", false);          if (temp == "not found") { temp = "0"; }
  64                         convert(temp, fromend);
  65
  66                         if ((end == 0) && (fromend == 0))  { m->mothurOut("You must provide either end or fromend for the chops.seqs command."); m->mothurOutEndLine(); abort = true;  }
  67                         if ((end != 0) && (fromend != 0))  { m->mothurOut("You must provide either end or fromend for the chops.seqs command, not both."); m->mothurOutEndLine(); abort = true;  }
  68                 }
  69
  70         }
  71         catch(exception& e) {
  72                 m->errorOut(e, "ChopSeqsCommand", "ChopSeqsCommand");
  73                 exit(1);
  74         }
  75 }
  76 //**********************************************************************************************************************
  77
  78 void ChopSeqsCommand::help(){
  79         try {
  80                 m->mothurOut("The chop.seqs command reads a fasta file and outputs a .chop.fasta with sequences trimmed to the end position.\n");
  81                 m->mothurOut("The chop.seqs command parameters are fasta, end and fromend, fasta is required.\n");
  82                 m->mothurOut("The chop.seqs command should be in the following format: chop.seqs(fasta=yourFasta, end=yourEnd).\n");
  83                 m->mothurOut("The end parameter allows you to specify an end base position for your sequences, default = 0.\n");
  84                 m->mothurOut("The fromend parameter allows you to remove the last X bases from the end of the sequence, default = 0.\n");
  85                 m->mothurOut("Example chop.seqs(fasta=amazon.fasta, end=200).\n");
  86                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  87         }
  88         catch(exception& e) {
  89                 m->errorOut(e, "ChopSeqsCommand", "help");
  90                 exit(1);
  91         }
  92 }
  93
  94 //**********************************************************************************************************************
  95
  96 int ChopSeqsCommand::execute(){
  97         try {
  98
  99                 if (abort == true) { return 0; }
 100
 101                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "chop.fasta";
 102                 string outputFileNameAccnos = outputDir + getRootName(getSimpleName(fastafile)) + "chop.accnos";
 103
 104                 ofstream out;
 105                 openOutputFile(outputFileName, out);
 106
 107                 ofstream outAcc;
 108                 openOutputFile(outputFileNameAccnos, outAcc);
 109
 110                 ifstream in;
 111                 openInputFile(fastafile, in);
 112
 113                 bool wroteAccnos = false;
 114
 115                 while (!in.eof()) {
 116
 117                         Sequence seq(in);
 118
 119                         if (m->control_pressed) { in.close(); out.close(); outAcc.close(); remove(outputFileName.c_str()); remove(outputFileNameAccnos.c_str()); return 0;  }
 120
 121                         if (seq.getName() != "") {
 122                                 string newSeqString = "";
 123                                 if (seq.getIsAligned()) { //sequence is aligned
 124                                         newSeqString = getChoppedAligned(seq);
 125                                 }else{
 126                                         newSeqString = getChoppedUnaligned(seq);
 127                                 }
 128
 129                                 //output trimmed sequence
 130                                 if (newSeqString != "") {
 131                                         out << ">" << seq.getName() << endl << newSeqString << endl;
 132                                 }else{
 133                                         outAcc << seq.getName() << endl;
 134                                         wroteAccnos = true;
 135                                 }
 136                         }
 137                 }
 138                 in.close();
 139                 out.close();
 140                 outAcc.close();
 141
 142                 m->mothurOutEndLine();
 143                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 144                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 145
 146                 if (wroteAccnos) { m->mothurOut(outputFileNameAccnos); m->mothurOutEndLine();  }
 147                 else {  remove(outputFileNameAccnos.c_str());  }
 148
 149                 m->mothurOutEndLine();
 150
 151                 return 0;
 152         }
 153
 154         catch(exception& e) {
 155                 m->errorOut(e, "ChopSeqsCommand", "execute");
 156                 exit(1);
 157         }
 158 }
 159 //**********************************************************************************************************************
 160 string ChopSeqsCommand::getChoppedUnaligned(Sequence seq) {
 161         try {
 162                 string temp = seq.getUnaligned();
 163
 164                 //if needed trim sequence
 165                 if (end != 0) {
 166                         if (temp.length() > end) {  temp = temp.substr(0, end);         }
 167                         else { temp = "";  }
 168                 }else { //you are using fromend
 169                         if (temp.length() > fromend) { temp = temp.substr(0, (temp.length()-fromend));  }
 170                         else {  temp = ""; } //sequence too short
 171                 }
 172
 173                 return temp;
 174         }
 175         catch(exception& e) {
 176                 m->errorOut(e, "ChopSeqsCommand", "getChoppedUnaligned");
 177                 exit(1);
 178         }
 179 }
 180 //**********************************************************************************************************************
 181 string ChopSeqsCommand::getChoppedAligned(Sequence seq) {
 182         try {
 183                 string temp = seq.getAligned();
 184                 string tempUnaligned = seq.getUnaligned();
 185
 186                 //if needed trim sequence
 187                 if (end != 0) {
 188                         if (tempUnaligned.length() > end) { //you have enough bases to remove some
 189
 190                                 int stopSpot = 0;
 191                                 int numBases = 0;
 192
 193                                 for (int i = 0; i < temp.length(); i++) {
 194                                         if(isalpha(temp[i])) { numBases++; }
 195
 196                                         if (numBases >= end) { stopSpot = i; break; }
 197                                 }
 198
 199                                 temp = temp.substr(0, stopSpot);
 200                         }else { temp = ""; } //sequence too short
 201
 202                 }else { //you are using fromend
 203
 204                         if (tempUnaligned.length() > fromend) {
 205
 206                                 int stopSpot = 0;
 207                                 int numBases = 0;
 208
 209                                 for (int i = (temp.length()-1); i >= 0; i--) {
 210                                         if(isalpha(temp[i])) { numBases++; }
 211
 212                                         if (numBases >= fromend) { stopSpot = i; break; }
 213                                 }
 214
 215                                 temp = temp.substr(0, stopSpot);
 216                         }
 217                         else {  temp = ""; } //sequence too short
 218                 }
 219
 220                 return temp;
 221         }
 222         catch(exception& e) {
 223                 m->errorOut(e, "ChopSeqsCommand", "getChoppedUnaligned");
 224                 exit(1);
 225         }
 226 }
 227 //**********************************************************************************************************************
 228
 229