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                 vector<int> pound;
 190                 vector<int> dash;
 191                 vector<int> plus;
 192                 vector<int> equal;
 193                 vector<int> loop;
 194                 vector<int> tilde;
 195                 vector<int> total;
 196
 197                 int count = 0;
 198                 while(!in.eof()){
 199                         if (m->control_pressed) { in.close(); out.close(); remove(outfile.c_str()); return 0; }
 200
 201                         Sequence seq(in);  m->gobble(in);
 202                         if (seq.getName() != "") {
 203                                 statData data = getStats(seq.getAligned());
 204
 205                                 if (haderror == 1) { m->control_pressed = true; break; }
 206
 207                                 int num = 1;
 208                                 if (namefile != "") {
 209                                         //make sure this sequence is in the namefile, else error
 210                                         map<string, int>::iterator it = nameMap.find(seq.getName());
 211
 212                                         if (it == nameMap.end()) { cout << "[ERROR]: " << seq.getName() << " is not in your namefile, please correct." << endl; m->control_pressed = true; }
 213                                         else { num = it->second; }
 214                                 }
 215
 216                                 //for each sequence this sequence represents
 217                                 for (int i = 0; i < num; i++) {
 218                                         pound.push_back(data.pound);
 219                                         dash.push_back(data.dash);
 220                                         plus.push_back(data.plus);
 221                                         equal.push_back(data.equal);
 222                                         loop.push_back(data.loop);
 223                                         tilde.push_back(data.tilde);
 224                                         total.push_back(data.total);
 225                                 }
 226                                 count++;
 227
 228                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 229                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << '\t' << num << endl;
 230                         }
 231                 }
 232
 233                 in.close();
 234                 out.close();
 235
 236                 if (m->control_pressed) {  remove(outfile.c_str()); return 0; }
 237
 238                 sort(pound.begin(), pound.end());
 239                 sort(dash.begin(), dash.end());
 240                 sort(plus.begin(), plus.end());
 241                 sort(equal.begin(), equal.end());
 242                 sort(loop.begin(), loop.end());
 243                 sort(tilde.begin(), tilde.end());
 244                 sort(total.begin(), total.end());
 245                 int size = pound.size();
 246
 247                 int ptile0_25   = int(size * 0.025);
 248                 int ptile25             = int(size * 0.250);
 249                 int ptile50             = int(size * 0.500);
 250                 int ptile75             = int(size * 0.750);
 251                 int ptile97_5   = int(size * 0.975);
 252                 int ptile100    = size - 1;
 253
 254                 if (m->control_pressed) {  remove(outfile.c_str()); return 0; }
 255
 256                 m->mothurOutEndLine();
 257                 m->mothurOut("\t\tPound\tDash\tPlus\tEqual\tLoop\tTilde\tTotal"); m->mothurOutEndLine();
 258                 m->mothurOut("Minimum:\t" + toString(pound[0]) + "\t" + toString(dash[0]) + "\t" + toString(plus[0]) + "\t" + toString(equal[0]) + "\t" + toString(loop[0]) + "\t" + toString(tilde[0]) + "\t" + toString(total[0])); m->mothurOutEndLine();
 259                 m->mothurOut("2.5%-tile:\t" + toString(pound[ptile0_25]) + "\t" + toString(dash[ptile0_25]) + "\t" + toString(plus[ptile0_25]) + "\t" + toString(equal[ptile0_25]) + "\t"+ toString(loop[ptile0_25]) + "\t"+ toString(tilde[ptile0_25]) + "\t"+ toString(total[ptile0_25])); m->mothurOutEndLine();
 260                 m->mothurOut("25%-tile:\t" + toString(pound[ptile25]) + "\t" + toString(dash[ptile25]) + "\t" + toString(plus[ptile25]) + "\t" + toString(equal[ptile25]) + "\t" + toString(loop[ptile25]) + "\t" + toString(tilde[ptile25]) + "\t" + toString(total[ptile25])); m->mothurOutEndLine();
 261                 m->mothurOut("Median: \t" + toString(pound[ptile50]) + "\t" + toString(dash[ptile50]) + "\t" + toString(plus[ptile50]) + "\t" + toString(equal[ptile50]) + "\t" + toString(loop[ptile50]) + "\t" + toString(tilde[ptile50]) + "\t" + toString(total[ptile50])); m->mothurOutEndLine();
 262                 m->mothurOut("75%-tile:\t" + toString(pound[ptile75]) + "\t" + toString(dash[ptile75]) + "\t" + toString(plus[ptile75]) + "\t" + toString(equal[ptile75]) + "\t" + toString(loop[ptile75]) + "\t" + toString(tilde[ptile75]) + "\t" + toString(total[ptile75])); m->mothurOutEndLine();
 263                 m->mothurOut("97.5%-tile:\t" + toString(pound[ptile97_5]) + "\t" + toString(dash[ptile97_5]) + "\t" + toString(plus[ptile97_5]) + "\t" + toString(equal[ptile97_5]) + "\t" + toString(loop[ptile97_5]) + "\t" + toString(tilde[ptile97_5]) + "\t" + toString(total[ptile97_5])); m->mothurOutEndLine();
 264                 m->mothurOut("Maximum:\t" + toString(pound[ptile100]) + "\t" + toString(dash[ptile100]) + "\t" + toString(plus[ptile100]) + "\t" + toString(equal[ptile100]) + "\t" + toString(loop[ptile100]) + "\t" + toString(tilde[ptile100]) + "\t" + toString(total[ptile100])); m->mothurOutEndLine();
 265                 if (namefile == "") {  m->mothurOut("# of Seqs:\t" + toString(count)); m->mothurOutEndLine(); }
 266                 else { m->mothurOut("# of unique seqs:\t" + toString(count)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(size)); m->mothurOutEndLine(); }
 267
 268
 269                 m->mothurOutEndLine();
 270                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 271                 m->mothurOut(outfile); m->mothurOutEndLine();   outputNames.push_back(outfile); outputTypes["aligncheck"].push_back(outfile);
 272                 m->mothurOutEndLine();
 273
 274                 return 0;
 275         }
 276
 277         catch(exception& e) {
 278                 m->errorOut(e, "AlignCheckCommand", "execute");
 279                 exit(1);
 280         }
 281 }
 282 //**********************************************************************************************************************
 283 void AlignCheckCommand::readMap(){
 284         try {
 285
 286                 structMap.resize(1, 0);
 287                 ifstream in;
 288
 289                 m->openInputFile(mapfile, in);
 290
 291                 while(!in.eof()){
 292                         int position;
 293                         in >> position;
 294                         structMap.push_back(position);
 295                         m->gobble(in);
 296                 }
 297                 in.close();
 298
 299                 seqLength = structMap.size();
 300
 301
 302                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 303                 for(int i=0;i<seqLength;i++){
 304                         if(structMap[i] != 0){
 305                                 if(structMap[structMap[i]] != i){
 306                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 307                                 }
 308                         }
 309                 }
 310
 311
 312         }
 313         catch(exception& e) {
 314                 m->errorOut(e, "AlignCheckCommand", "readMap");
 315                 exit(1);
 316         }
 317 }
 318 /**************************************************************************************************/
 319
 320 statData AlignCheckCommand::getStats(string sequence){
 321         try {
 322
 323                 statData data;
 324                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 325
 326                 int length = sequence.length();
 327
 328                 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;  }
 329
 330                 for(int i=1;i<length;i++){
 331                         if(structMap[i] != 0){
 332                                 if(sequence[i] == 'A'){
 333                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 334                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 335                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 336                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 337                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 338                                         data.total++;
 339                                 }
 340                                 else if(sequence[i] == 'T'){
 341                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 342                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 343                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 344                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 345                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 346                                         data.total++;
 347                                 }
 348                                 else if(sequence[i] == 'G'){
 349                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 350                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 351                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 352                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 353                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 354                                         data.total++;
 355                                 }
 356                                 else if(sequence[i] == 'C'){
 357                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 358                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 359                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 360                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 361                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 362                                         data.total++;
 363                                 }
 364                                 else if(sequence[i] == '-'){
 365                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 366                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 367                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 368                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 369                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 370                                 }
 371                         }
 372                         else if(isalnum(sequence[i])){
 373                                 data.loop++;
 374                                 data.total++;
 375                         }
 376                 }
 377                 return data;
 378
 379         }
 380         catch(exception& e) {
 381                 m->errorOut(e, "AlignCheckCommand", "getStats");
 382                 exit(1);
 383         }
 384 }
 385 //**********************************************************************************************************************