listseqscommand.cpp

   1 /*
   2  *  listseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "listseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14
  15 ListSeqsCommand::ListSeqsCommand(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","name", "group", "alignreport" };
  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
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //check for required parameters
  38                         fastafile = validParameter.validFile(parameters, "fasta", true);
  39                         if (fastafile == "not open") { abort = true; }
  40                         else if (fastafile == "not found") {  fastafile = "";  }
  41
  42                         namefile = validParameter.validFile(parameters, "name", true);
  43                         if (namefile == "not open") { abort = true; }
  44                         else if (namefile == "not found") {  namefile = "";  }
  45
  46                         groupfile = validParameter.validFile(parameters, "group", true);
  47                         if (groupfile == "not open") { abort = true; }
  48                         else if (groupfile == "not found") {  groupfile = "";  }
  49
  50                         alignfile = validParameter.validFile(parameters, "alignreport", true);
  51                         if (alignfile == "not open") { abort = true; }
  52                         else if (alignfile == "not found") {  alignfile = "";  }
  53
  54                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == ""))  { mothurOut("You must provide a file."); mothurOutEndLine(); abort = true; }
  55
  56                         if (parameters.size() > 1) { mothurOut("You may only enter one file."); mothurOutEndLine(); abort = true;  }
  57                 }
  58
  59         }
  60         catch(exception& e) {
  61                 errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66
  67 void ListSeqsCommand::help(){
  68         try {
  69                 mothurOut("The list.seqs command reads a fasta, name, group or alignreport file and outputs a .accnos file containing sequence names.\n");
  70                 mothurOut("The list.seqs command parameters are fasta, name, group and alignreport.  You must provide one of these parameters.\n");
  71                 mothurOut("The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n");
  72                 mothurOut("Example list.seqs(fasta=amazon.fasta).\n");
  73                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  74         }
  75         catch(exception& e) {
  76                 errorOut(e, "ListSeqsCommand", "help");
  77                 exit(1);
  78         }
  79 }
  80
  81 //**********************************************************************************************************************
  82
  83 int ListSeqsCommand::execute(){
  84         try {
  85
  86                 if (abort == true) { return 0; }
  87
  88                 //read functions fill names vector
  89                 if (fastafile != "")            {       inputFileName = fastafile;      readFasta();    }
  90                 else if (namefile != "")        {       inputFileName = namefile;       readName();             }
  91                 else if (groupfile != "")       {       inputFileName = groupfile;      readGroup();    }
  92                 else if (alignfile != "")       {       inputFileName = alignfile;      readAlign();    }
  93
  94                 //sort in alphabetical order
  95                 sort(names.begin(), names.end());
  96
  97                 string outputFileName = getRootName(inputFileName) + "accnos";
  98                 ofstream out;
  99                 openOutputFile(outputFileName, out);
 100
 101                 //output to .accnos file
 102                 for (int i = 0; i < names.size(); i++) {
 103                         out << names[i] << endl;
 104                 }
 105                 out.close();
 106
 107                 return 0;
 108         }
 109
 110         catch(exception& e) {
 111                 errorOut(e, "ListSeqsCommand", "execute");
 112                 exit(1);
 113         }
 114 }
 115
 116 //**********************************************************************************************************************
 117 void ListSeqsCommand::readFasta(){
 118         try {
 119
 120                 ifstream in;
 121                 openInputFile(fastafile, in);
 122                 string name;
 123
 124                 while(!in.eof()){
 125                         Sequence currSeq(in);
 126                         name = currSeq.getName();
 127
 128                         names.push_back(name);
 129
 130                         gobble(in);
 131                 }
 132                 in.close();
 133
 134         }
 135         catch(exception& e) {
 136                 errorOut(e, "ListSeqsCommand", "readFasta");
 137                 exit(1);
 138         }
 139 }
 140
 141 //**********************************************************************************************************************
 142 void ListSeqsCommand::readName(){
 143         try {
 144
 145                 ifstream in;
 146                 openInputFile(namefile, in);
 147                 string name, firstCol, secondCol;
 148
 149                 while(!in.eof()){
 150
 151                         in >> firstCol;
 152                         in >> secondCol;
 153
 154                         //parse second column saving each name
 155                         while (secondCol.find_first_of(',') != -1) {
 156                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 157                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 158                                 names.push_back(name);
 159                         }
 160
 161                         //get name after last ,
 162                         names.push_back(secondCol);
 163
 164                         gobble(in);
 165                 }
 166                 in.close();
 167
 168         }
 169         catch(exception& e) {
 170                 errorOut(e, "ListSeqsCommand", "readName");
 171                 exit(1);
 172         }
 173 }
 174
 175 //**********************************************************************************************************************
 176 void ListSeqsCommand::readGroup(){
 177         try {
 178
 179                 ifstream in;
 180                 openInputFile(groupfile, in);
 181                 string name, group;
 182
 183                 while(!in.eof()){
 184
 185                         in >> name;                             //read from first column
 186                         in >> group;                    //read from second column
 187
 188                         names.push_back(name);
 189
 190                         gobble(in);
 191                 }
 192                 in.close();
 193
 194         }
 195         catch(exception& e) {
 196                 errorOut(e, "ListSeqsCommand", "readGroup");
 197                 exit(1);
 198         }
 199 }
 200
 201 //**********************************************************************************************************************
 202 //alignreport file has a column header line then all other lines contain 16 columns.  we just want the first column since that contains the name
 203 void ListSeqsCommand::readAlign(){
 204         try {
 205
 206                 ifstream in;
 207                 openInputFile(alignfile, in);
 208                 string name, junk;
 209
 210                 //read column headers
 211                 for (int i = 0; i < 16; i++) {
 212                         if (!in.eof())  {       in >> junk;             }
 213                         else                    {       break;                  }
 214                 }
 215
 216
 217                 while(!in.eof()){
 218
 219                         in >> name;                             //read from first column
 220
 221                         //read rest
 222                         for (int i = 0; i < 15; i++) {
 223                                 if (!in.eof())  {       in >> junk;             }
 224                                 else                    {       break;                  }
 225                         }
 226
 227                         names.push_back(name);
 228
 229                         gobble(in);
 230                 }
 231                 in.close();
 232
 233
 234         }
 235         catch(exception& e) {
 236                 errorOut(e, "ListSeqsCommand", "readAlign");
 237                 exit(1);
 238         }
 239 }
 240 //**********************************************************************************************************************