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