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", "name","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                 abort = true; calledHelp = true;
  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; calledHelp = false;
  65                 haderror = 0;
  66
  67                 //allow user to run help
  68                 if(option == "help") { help(); abort = true; calledHelp = true; }
  69
  70                 else {
  71                         //valid paramters for this command
  72                         string Array[] =  {"fasta","name","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                                 it = parameters.find("name");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 117                                 }
 118                         }
 119
 120                         //check for required parameters
 121                         mapfile = validParameter.validFile(parameters, "map", true);
 122                         if (mapfile == "not open") { abort = true; }
 123                         else if (mapfile == "not found") {  mapfile = "";  m->mothurOut("You must provide an map file."); m->mothurOutEndLine(); abort = true; }
 124
 125                         fastafile = validParameter.validFile(parameters, "fasta", true);
 126                         if (fastafile == "not open") { abort = true; }
 127                         else if (fastafile == "not found") {  fastafile = "";  m->mothurOut("You must provide an fasta file."); m->mothurOutEndLine(); abort = true;  }
 128
 129                         namefile = validParameter.validFile(parameters, "name", true);
 130                         if (namefile == "not open") { namefile = ""; abort = true; }
 131                         else if (namefile == "not found") { namefile = "";  }
 132
 133                         //if the user changes the output directory command factory will send this info to us in the output parameter
 134                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 135                                 outputDir = "";
 136                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 137                         }
 138
 139                 }
 140
 141         }
 142         catch(exception& e) {
 143                 m->errorOut(e, "AlignCheckCommand", "RemoveSeqsCommand");
 144                 exit(1);
 145         }
 146 }
 147 //**********************************************************************************************************************
 148
 149 void AlignCheckCommand::help(){
 150         try {
 151                 m->mothurOut("The align.check command reads a fasta file and map file.\n");
 152                 m->mothurOut("It outputs a file containing the secondary structure matches in the .align.check file.\n");
 153                 m->mothurOut("The align.check command parameters are fasta and map, both are required.\n");
 154                 m->mothurOut("The align.check command should be in the following format: align.check(fasta=yourFasta, map=yourMap).\n");
 155                 m->mothurOut("Example align.check(map=silva.ss.map, fasta=amazon.fasta).\n");
 156                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 157         }
 158         catch(exception& e) {
 159                 m->errorOut(e, "AlignCheckCommand", "help");
 160                 exit(1);
 161         }
 162 }
 163
 164 //**********************************************************************************************************************
 165
 166 int AlignCheckCommand::execute(){
 167         try {
 168
 169                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 170
 171                 //get secondary structure info.
 172                 readMap();
 173
 174                 if (namefile != "") { nameMap = m->readNames(namefile); }
 175
 176                 if (m->control_pressed) { return 0; }
 177
 178                 ifstream in;
 179                 m->openInputFile(fastafile, in);
 180
 181                 ofstream out;
 182                 string outfile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "align.check";
 183                 m->openOutputFile(outfile, out);
 184
 185
 186                 out << "name" << '\t' << "pound" << '\t' << "dash" << '\t' << "plus" << '\t' << "equal" << '\t';
 187                 out << "loop" << '\t' << "tilde" << '\t' << "total"  << '\t' << "numseqs" << endl;
 188
 189
 190                 while(!in.eof()){
 191                         if (m->control_pressed) { in.close(); out.close(); remove(outfile.c_str()); return 0; }
 192
 193                         Sequence seq(in);  m->gobble(in);
 194                         if (seq.getName() != "") {
 195                                 statData data = getStats(seq.getAligned());
 196
 197                                 if (haderror == 1) { break; }
 198
 199                                 int num = 1;
 200                                 if (namefile != "") {
 201                                         //make sure this sequence is in the namefile, else error
 202                                         map<string, int>::iterator it = nameMap.find(seq.getName());
 203
 204                                         if (it == nameMap.end()) { cout << "[ERROR]: " << seq.getName() << " is not in your namefile, please correct." << endl; m->control_pressed = true; }
 205                                         else { num = it->second; }
 206                                 }
 207
 208                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 209                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << '\t' << num << endl;
 210                         }
 211                 }
 212
 213                 in.close();
 214                 out.close();
 215
 216                 if (m->control_pressed) {  remove(outfile.c_str()); return 0; }
 217
 218                 m->mothurOutEndLine();
 219                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 220                 m->mothurOut(outfile); m->mothurOutEndLine();   outputNames.push_back(outfile); outputTypes["aligncheck"].push_back(outfile);
 221                 m->mothurOutEndLine();
 222
 223                 return 0;
 224         }
 225
 226         catch(exception& e) {
 227                 m->errorOut(e, "AlignCheckCommand", "execute");
 228                 exit(1);
 229         }
 230 }
 231 //**********************************************************************************************************************
 232 void AlignCheckCommand::readMap(){
 233         try {
 234
 235                 structMap.resize(1, 0);
 236                 ifstream in;
 237
 238                 m->openInputFile(mapfile, in);
 239
 240                 while(!in.eof()){
 241                         int position;
 242                         in >> position;
 243                         structMap.push_back(position);
 244                         m->gobble(in);
 245                 }
 246                 in.close();
 247
 248                 seqLength = structMap.size();
 249
 250
 251                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 252                 for(int i=0;i<seqLength;i++){
 253                         if(structMap[i] != 0){
 254                                 if(structMap[structMap[i]] != i){
 255                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 256                                 }
 257                         }
 258                 }
 259
 260
 261         }
 262         catch(exception& e) {
 263                 m->errorOut(e, "AlignCheckCommand", "readMap");
 264                 exit(1);
 265         }
 266 }
 267 /**************************************************************************************************/
 268
 269 statData AlignCheckCommand::getStats(string sequence){
 270         try {
 271
 272                 statData data;
 273                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 274
 275                 int length = sequence.length();
 276
 277                 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;  }
 278
 279                 for(int i=1;i<length;i++){
 280                         if(structMap[i] != 0){
 281                                 if(sequence[i] == 'A'){
 282                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 283                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 284                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 285                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 286                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 287                                         data.total++;
 288                                 }
 289                                 else if(sequence[i] == 'T'){
 290                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 291                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 292                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 293                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 294                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 295                                         data.total++;
 296                                 }
 297                                 else if(sequence[i] == 'G'){
 298                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 299                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 300                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 301                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 302                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 303                                         data.total++;
 304                                 }
 305                                 else if(sequence[i] == 'C'){
 306                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 307                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 308                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 309                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 310                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 311                                         data.total++;
 312                                 }
 313                                 else if(sequence[i] == '-'){
 314                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 315                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 316                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 317                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 318                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 319                                 }
 320                         }
 321                         else if(isalnum(sequence[i])){
 322                                 data.loop++;
 323                                 data.total++;
 324                         }
 325                 }
 326                 return data;
 327
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "AlignCheckCommand", "getStats");
 331                 exit(1);
 332         }
 333 }
 334 //**********************************************************************************************************************