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