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::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pmap("map", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pmap);
  18         CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  20                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  21
  22                 vector<string> myArray;
  23                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  24                 return myArray;
  25         }
  26         catch(exception& e) {
  27                 m->errorOut(e, "AlignCheckCommand", "setParameters");
  28                 exit(1);
  29         }
  30 }
  31 //**********************************************************************************************************************
  32 string AlignCheckCommand::getHelpString(){
  33         try {
  34                 string helpString = "";
  35                 helpString += "The align.check command reads a fasta file and map file as well as an optional name file.\n";
  36                 helpString += "It outputs a file containing the secondary structure matches in the .align.check file.\n";
  37                 helpString += "The align.check command parameters are fasta and map, both are required.\n";
  38                 helpString += "The align.check command should be in the following format: align.check(fasta=yourFasta, map=yourMap).\n";
  39                 helpString += "Example align.check(map=silva.ss.map, fasta=amazon.fasta).\n";
  40                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  41                 return helpString;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "AlignCheckCommand", "getHelpString");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 string AlignCheckCommand::getOutputFileNameTag(string type, string inputName=""){
  50         try {
  51         string outputFileName = "";
  52                 map<string, vector<string> >::iterator it;
  53
  54         //is this a type this command creates
  55         it = outputTypes.find(type);
  56         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  57         else {
  58             if (type == "aligncheck")            {   outputFileName =  "align.check";   }
  59             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  60         }
  61         return outputFileName;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "AlignCheckCommand", "getOutputFileNameTag");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 AlignCheckCommand::AlignCheckCommand(){
  70         try {
  71                 abort = true; calledHelp = true;
  72                 setParameters();
  73                 vector<string> tempOutNames;
  74                 outputTypes["aligncheck"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82
  83 AlignCheckCommand::AlignCheckCommand(string option)  {
  84         try {
  85                 abort = false; calledHelp = false;
  86                 haderror = 0;
  87
  88                 //allow user to run help
  89                 if(option == "help") { help(); abort = true; calledHelp = true; }
  90                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  91
  92                 else {
  93                         vector<string> myArray = setParameters();
  94
  95                         OptionParser parser(option);
  96                         map<string,string> parameters = parser.getParameters();
  97
  98                         ValidParameters validParameter;
  99                         map<string,string>::iterator it;
 100
 101                         //check to make sure all parameters are valid for command
 102                         for (it = parameters.begin(); it != parameters.end(); it++) {
 103                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 104                         }
 105
 106                         //initialize outputTypes
 107                         vector<string> tempOutNames;
 108                         outputTypes["aligncheck"] = tempOutNames;
 109
 110                         //if the user changes the input directory command factory will send this info to us in the output parameter
 111                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 112                         if (inputDir == "not found"){   inputDir = "";          }
 113                         else {
 114                                 string path;
 115                                 it = parameters.find("fasta");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 121                                 }
 122
 123                                 it = parameters.find("map");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["map"] = inputDir + it->second;              }
 129                                 }
 130
 131                                 it = parameters.find("name");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 137                                 }
 138                         }
 139
 140                         //check for required parameters
 141                         mapfile = validParameter.validFile(parameters, "map", true);
 142                         if (mapfile == "not open") { abort = true; }
 143                         else if (mapfile == "not found") {  mapfile = "";  m->mothurOut("You must provide an map file."); m->mothurOutEndLine(); abort = true; }
 144
 145                         fastafile = validParameter.validFile(parameters, "fasta", true);
 146                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 147                         else if (fastafile == "not found") {
 148                                 fastafile = m->getFastaFile();
 149                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 150                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 151                         }else { m->setFastaFile(fastafile); }
 152
 153                         namefile = validParameter.validFile(parameters, "name", true);
 154                         if (namefile == "not open") { namefile = ""; abort = true; }
 155                         else if (namefile == "not found") { namefile = "";  }
 156                         else { m->setNameFile(namefile); }
 157
 158                         //if the user changes the output directory command factory will send this info to us in the output parameter
 159                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 160                                 outputDir = "";
 161                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 162                         }
 163
 164                         if ((namefile == "") && (fastafile != "")){
 165                                 vector<string> files; files.push_back(fastafile);
 166                                 parser.getNameFile(files);
 167                         }
 168                 }
 169
 170         }
 171         catch(exception& e) {
 172                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
 173                 exit(1);
 174         }
 175 }
 176 //**********************************************************************************************************************
 177
 178 int AlignCheckCommand::execute(){
 179         try {
 180
 181                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 182
 183                 //get secondary structure info.
 184                 readMap();
 185
 186                 if (namefile != "") { nameMap = m->readNames(namefile); }
 187
 188                 if (m->control_pressed) { return 0; }
 189
 190                 ifstream in;
 191                 m->openInputFile(fastafile, in);
 192
 193                 ofstream out;
 194                 string outfile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("aligncheck");
 195                 m->openOutputFile(outfile, out);
 196
 197
 198                 out << "name" << '\t' << "pound" << '\t' << "dash" << '\t' << "plus" << '\t' << "equal" << '\t';
 199                 out << "loop" << '\t' << "tilde" << '\t' << "total"  << '\t' << "numseqs" << endl;
 200
 201                 vector<int> pound;
 202                 vector<int> dash;
 203                 vector<int> plus;
 204                 vector<int> equal;
 205                 vector<int> loop;
 206                 vector<int> tilde;
 207                 vector<int> total;
 208
 209                 int count = 0;
 210                 while(!in.eof()){
 211                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outfile); return 0; }
 212
 213                         Sequence seq(in);  m->gobble(in);
 214                         if (seq.getName() != "") {
 215                                 statData data = getStats(seq.getAligned());
 216
 217                                 if (haderror == 1) { m->control_pressed = true; break; }
 218
 219                                 int num = 1;
 220                                 if (namefile != "") {
 221                                         //make sure this sequence is in the namefile, else error
 222                                         map<string, int>::iterator it = nameMap.find(seq.getName());
 223
 224                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + seq.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 225                                         else { num = it->second; }
 226                                 }
 227
 228                                 //for each sequence this sequence represents
 229                                 for (int i = 0; i < num; i++) {
 230                                         pound.push_back(data.pound);
 231                                         dash.push_back(data.dash);
 232                                         plus.push_back(data.plus);
 233                                         equal.push_back(data.equal);
 234                                         loop.push_back(data.loop);
 235                                         tilde.push_back(data.tilde);
 236                                         total.push_back(data.total);
 237                                 }
 238                                 count++;
 239
 240                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 241                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << '\t' << num << endl;
 242                         }
 243                 }
 244
 245                 in.close();
 246                 out.close();
 247
 248                 if (m->control_pressed) {  m->mothurRemove(outfile); return 0; }
 249
 250                 sort(pound.begin(), pound.end());
 251                 sort(dash.begin(), dash.end());
 252                 sort(plus.begin(), plus.end());
 253                 sort(equal.begin(), equal.end());
 254                 sort(loop.begin(), loop.end());
 255                 sort(tilde.begin(), tilde.end());
 256                 sort(total.begin(), total.end());
 257                 int size = pound.size();
 258
 259                 int ptile0_25   = int(size * 0.025);
 260                 int ptile25             = int(size * 0.250);
 261                 int ptile50             = int(size * 0.500);
 262                 int ptile75             = int(size * 0.750);
 263                 int ptile97_5   = int(size * 0.975);
 264                 int ptile100    = size - 1;
 265
 266                 if (m->control_pressed) {  m->mothurRemove(outfile); return 0; }
 267
 268                 m->mothurOutEndLine();
 269                 m->mothurOut("\t\tPound\tDash\tPlus\tEqual\tLoop\tTilde\tTotal"); m->mothurOutEndLine();
 270                 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();
 271                 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();
 272                 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();
 273                 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();
 274                 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();
 275                 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();
 276                 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();
 277                 if (namefile == "") {  m->mothurOut("# of Seqs:\t" + toString(count)); m->mothurOutEndLine(); }
 278                 else { m->mothurOut("# of unique seqs:\t" + toString(count)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(size)); m->mothurOutEndLine(); }
 279
 280
 281                 m->mothurOutEndLine();
 282                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 283                 m->mothurOut(outfile); m->mothurOutEndLine();   outputNames.push_back(outfile); outputTypes["aligncheck"].push_back(outfile);
 284                 m->mothurOutEndLine();
 285
 286                 return 0;
 287         }
 288
 289         catch(exception& e) {
 290                 m->errorOut(e, "AlignCheckCommand", "execute");
 291                 exit(1);
 292         }
 293 }
 294 //**********************************************************************************************************************
 295 void AlignCheckCommand::readMap(){
 296         try {
 297
 298                 structMap.resize(1, 0);
 299                 ifstream in;
 300
 301                 m->openInputFile(mapfile, in);
 302
 303                 while(!in.eof()){
 304                         int position;
 305                         in >> position;
 306                         structMap.push_back(position);
 307                         m->gobble(in);
 308                 }
 309                 in.close();
 310
 311                 seqLength = structMap.size();
 312
 313
 314                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 315                 for(int i=0;i<seqLength;i++){
 316                         if(structMap[i] != 0){
 317                                 if(structMap[structMap[i]] != i){
 318                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 319                                 }
 320                         }
 321                 }
 322
 323
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "AlignCheckCommand", "readMap");
 327                 exit(1);
 328         }
 329 }
 330 /**************************************************************************************************/
 331
 332 statData AlignCheckCommand::getStats(string sequence){
 333         try {
 334
 335                 statData data;
 336                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 337
 338                 int length = sequence.length();
 339
 340                 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;  }
 341
 342                 for(int i=1;i<length;i++){
 343                         if(structMap[i] != 0){
 344                                 if(sequence[i] == 'A'){
 345                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 346                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 347                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 348                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 349                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 350                                         data.total++;
 351                                 }
 352                                 else if(sequence[i] == 'T'){
 353                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 354                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 355                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 356                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 357                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 358                                         data.total++;
 359                                 }
 360                                 else if(sequence[i] == 'G'){
 361                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 362                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 363                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 364                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 365                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 366                                         data.total++;
 367                                 }
 368                                 else if(sequence[i] == 'C'){
 369                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 370                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 371                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 372                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 373                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 374                                         data.total++;
 375                                 }
 376                                 else if(sequence[i] == '-'){
 377                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 378                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 379                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 380                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 381                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 382                                 }
 383                         }
 384                         else if(isalnum(sequence[i])){
 385                                 data.loop++;
 386                                 data.total++;
 387                         }
 388                 }
 389                 return data;
 390
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "AlignCheckCommand", "getStats");
 394                 exit(1);
 395         }
 396 }
 397 //**********************************************************************************************************************