secondarystructurecommand.cpp

   1 /*
   2  *  secondarystructurecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/18/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "secondarystructurecommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> AlignCheckCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta","map", "outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "AlignCheckCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 AlignCheckCommand::AlignCheckCommand(){
  27         try {
  28                 //initialize outputTypes
  29                 vector<string> tempOutNames;
  30                 outputTypes["aligncheck"] = tempOutNames;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 vector<string> AlignCheckCommand::getRequiredParameters(){
  39         try {
  40                 string Array[] =  {"fasta","map"};
  41                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  42                 return myArray;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "AlignCheckCommand", "getRequiredParameters");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 vector<string> AlignCheckCommand::getRequiredFiles(){
  51         try {
  52                 vector<string> myArray;
  53                 return myArray;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "AlignCheckCommand", "getRequiredFiles");
  57                 exit(1);
  58         }
  59 }
  60 //**********************************************************************************************************************
  61
  62 AlignCheckCommand::AlignCheckCommand(string option)  {
  63         try {
  64                 abort = false;
  65                 haderror = 0;
  66
  67                 //allow user to run help
  68                 if(option == "help") { help(); abort = true; }
  69
  70                 else {
  71                         //valid paramters for this command
  72                         string Array[] =  {"fasta","map", "outputdir","inputdir"};
  73                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  74
  75                         OptionParser parser(option);
  76                         map<string,string> parameters = parser.getParameters();
  77
  78                         ValidParameters validParameter;
  79                         map<string,string>::iterator it;
  80
  81                         //check to make sure all parameters are valid for command
  82                         for (it = parameters.begin(); it != parameters.end(); it++) {
  83                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  84                         }
  85
  86                         //initialize outputTypes
  87                         vector<string> tempOutNames;
  88                         outputTypes["aligncheck"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("fasta");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 101                                 }
 102
 103                                 it = parameters.find("map");
 104                                 //user has given a template file
 105                                 if(it != parameters.end()){
 106                                         path = m->hasPath(it->second);
 107                                         //if the user has not given a path then, add inputdir. else leave path alone.
 108                                         if (path == "") {       parameters["map"] = inputDir + it->second;              }
 109                                 }
 110                         }
 111
 112                         //check for required parameters
 113                         mapfile = validParameter.validFile(parameters, "map", true);
 114                         if (mapfile == "not open") { abort = true; }
 115                         else if (mapfile == "not found") {  mapfile = "";  m->mothurOut("You must provide an map file."); m->mothurOutEndLine(); abort = true; }
 116
 117                         fastafile = validParameter.validFile(parameters, "fasta", true);
 118                         if (fastafile == "not open") { abort = true; }
 119                         else if (fastafile == "not found") {  fastafile = "";  m->mothurOut("You must provide an fasta file."); m->mothurOutEndLine(); abort = true;  }
 120
 121                         //if the user changes the output directory command factory will send this info to us in the output parameter
 122                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 123                                 outputDir = "";
 124                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 125                         }
 126
 127                 }
 128
 129         }
 130         catch(exception& e) {
 131                 m->errorOut(e, "AlignCheckCommand", "RemoveSeqsCommand");
 132                 exit(1);
 133         }
 134 }
 135 //**********************************************************************************************************************
 136
 137 void AlignCheckCommand::help(){
 138         try {
 139                 m->mothurOut("The align.check command reads a fasta file and map file.\n");
 140                 m->mothurOut("It outputs a file containing the secondary structure matches in the .align.check file.\n");
 141                 m->mothurOut("The align.check command parameters are fasta and map, both are required.\n");
 142                 m->mothurOut("The align.check command should be in the following format: align.check(fasta=yourFasta, map=yourMap).\n");
 143                 m->mothurOut("Example align.check(map=silva.ss.map, fasta=amazon.fasta).\n");
 144                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 145         }
 146         catch(exception& e) {
 147                 m->errorOut(e, "AlignCheckCommand", "help");
 148                 exit(1);
 149         }
 150 }
 151
 152 //**********************************************************************************************************************
 153
 154 int AlignCheckCommand::execute(){
 155         try {
 156
 157                 if (abort == true) { return 0; }
 158
 159                 //get secondary structure info.
 160                 readMap();
 161
 162                 ifstream in;
 163                 m->openInputFile(fastafile, in);
 164
 165                 ofstream out;
 166                 string outfile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "align.check";
 167                 m->openOutputFile(outfile, out);
 168
 169                 out << "name" << '\t' << "pound" << '\t' << "dash" << '\t' << "plus" << '\t' << "equal" << '\t';
 170                 out << "loop" << '\t' << "tilde" << '\t' << "total" << endl;
 171
 172
 173                 while(!in.eof()){
 174                         if (m->control_pressed) { in.close(); out.close(); remove(outfile.c_str()); return 0; }
 175
 176                         Sequence seq(in);  m->gobble(in);
 177                         if (seq.getName() != "") {
 178                                 statData data = getStats(seq.getAligned());
 179
 180                                 if (haderror == 1) { break; }
 181
 182                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 183                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << endl;
 184                         }
 185                 }
 186
 187                 in.close();
 188                 out.close();
 189
 190                 if (m->control_pressed) {  remove(outfile.c_str()); return 0; }
 191
 192                 m->mothurOutEndLine();
 193                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 194                 m->mothurOut(outfile); m->mothurOutEndLine();   outputNames.push_back(outfile); outputTypes["aligncheck"].push_back(outfile);
 195                 m->mothurOutEndLine();
 196
 197                 return 0;
 198         }
 199
 200         catch(exception& e) {
 201                 m->errorOut(e, "AlignCheckCommand", "execute");
 202                 exit(1);
 203         }
 204 }
 205 //**********************************************************************************************************************
 206 void AlignCheckCommand::readMap(){
 207         try {
 208
 209                 structMap.resize(1, 0);
 210                 ifstream in;
 211
 212                 m->openInputFile(mapfile, in);
 213
 214                 while(!in.eof()){
 215                         int position;
 216                         in >> position;
 217                         structMap.push_back(position);
 218                         m->gobble(in);
 219                 }
 220                 in.close();
 221
 222                 seqLength = structMap.size();
 223
 224
 225                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 226                 for(int i=0;i<seqLength;i++){
 227                         if(structMap[i] != 0){
 228                                 if(structMap[structMap[i]] != i){
 229                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 230                                 }
 231                         }
 232                 }
 233
 234
 235         }
 236         catch(exception& e) {
 237                 m->errorOut(e, "AlignCheckCommand", "readMap");
 238                 exit(1);
 239         }
 240 }
 241 /**************************************************************************************************/
 242
 243 statData AlignCheckCommand::getStats(string sequence){
 244         try {
 245
 246                 statData data;
 247                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 248
 249                 int length = sequence.length();
 250
 251                 if (length != seqLength) { m->mothurOut("your sequences are " + toString(length) + " long, but your map file only contains " + toString(seqLength) + " entries. please correct."); m->mothurOutEndLine(); haderror = 1; return data;  }
 252
 253                 for(int i=1;i<length;i++){
 254                         if(structMap[i] != 0){
 255                                 if(sequence[i] == 'A'){
 256                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 257                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 258                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 259                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 260                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 261                                         data.total++;
 262                                 }
 263                                 else if(sequence[i] == 'T'){
 264                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 265                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 266                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 267                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 268                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 269                                         data.total++;
 270                                 }
 271                                 else if(sequence[i] == 'G'){
 272                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 273                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 274                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 275                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 276                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 277                                         data.total++;
 278                                 }
 279                                 else if(sequence[i] == 'C'){
 280                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 281                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 282                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 283                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 284                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 285                                         data.total++;
 286                                 }
 287                                 else if(sequence[i] == '-'){
 288                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 289                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 290                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 291                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 292                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 293                                 }
 294                         }
 295                         else if(isalnum(sequence[i])){
 296                                 data.loop++;
 297                                 data.total++;
 298                         }
 299                 }
 300                 return data;
 301
 302         }
 303         catch(exception& e) {
 304                 m->errorOut(e, "AlignCheckCommand", "getStats");
 305                 exit(1);
 306         }
 307 }
 308
 309
 310 //**********************************************************************************************************************