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 #include "counttable.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> AlignCheckCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pmap("map", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pmap);
  19                 CommandParameter pname("name", "InputTypes", "", "", "namecount", "none", "none",false,false); parameters.push_back(pname);
  20         CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none",false,false); parameters.push_back(pcount);
  21         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "AlignCheckCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string AlignCheckCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The align.check command reads a fasta file and map file as well as an optional name or count file.\n";
  38                 helpString += "It outputs a file containing the secondary structure matches in the .align.check file.\n";
  39                 helpString += "The align.check command parameters are fasta and map, both are required.\n";
  40                 helpString += "The align.check command should be in the following format: align.check(fasta=yourFasta, map=yourMap).\n";
  41                 helpString += "Example align.check(map=silva.ss.map, fasta=amazon.fasta).\n";
  42                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  43                 return helpString;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "AlignCheckCommand", "getHelpString");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 string AlignCheckCommand::getOutputFileNameTag(string type, string inputName=""){
  52         try {
  53         string outputFileName = "";
  54                 map<string, vector<string> >::iterator it;
  55
  56         //is this a type this command creates
  57         it = outputTypes.find(type);
  58         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  59         else {
  60             if (type == "aligncheck")            {   outputFileName =  "align.check";   }
  61             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  62         }
  63         return outputFileName;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "AlignCheckCommand", "getOutputFileNameTag");
  67                 exit(1);
  68         }
  69 }
  70 //**********************************************************************************************************************
  71 AlignCheckCommand::AlignCheckCommand(){
  72         try {
  73                 abort = true; calledHelp = true;
  74                 setParameters();
  75                 vector<string> tempOutNames;
  76                 outputTypes["aligncheck"] = tempOutNames;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84
  85 AlignCheckCommand::AlignCheckCommand(string option)  {
  86         try {
  87                 abort = false; calledHelp = false;
  88                 haderror = 0;
  89
  90                 //allow user to run help
  91                 if(option == "help") { help(); abort = true; calledHelp = true; }
  92                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  93
  94                 else {
  95                         vector<string> myArray = setParameters();
  96
  97                         OptionParser parser(option);
  98                         map<string,string> parameters = parser.getParameters();
  99
 100                         ValidParameters validParameter;
 101                         map<string,string>::iterator it;
 102
 103                         //check to make sure all parameters are valid for command
 104                         for (it = parameters.begin(); it != parameters.end(); it++) {
 105                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["aligncheck"] = tempOutNames;
 111
 112                         //if the user changes the input directory command factory will send this info to us in the output parameter
 113                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 114                         if (inputDir == "not found"){   inputDir = "";          }
 115                         else {
 116                                 string path;
 117                                 it = parameters.find("fasta");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 123                                 }
 124
 125                                 it = parameters.find("map");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["map"] = inputDir + it->second;              }
 131                                 }
 132
 133                                 it = parameters.find("name");
 134                                 //user has given a template file
 135                                 if(it != parameters.end()){
 136                                         path = m->hasPath(it->second);
 137                                         //if the user has not given a path then, add inputdir. else leave path alone.
 138                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 139                                 }
 140
 141                 it = parameters.find("count");
 142                                 //user has given a template file
 143                                 if(it != parameters.end()){
 144                                         path = m->hasPath(it->second);
 145                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 147                                 }
 148                         }
 149
 150                         //check for required parameters
 151                         mapfile = validParameter.validFile(parameters, "map", true);
 152                         if (mapfile == "not open") { abort = true; }
 153                         else if (mapfile == "not found") {  mapfile = "";  m->mothurOut("You must provide an map file."); m->mothurOutEndLine(); abort = true; }
 154
 155                         fastafile = validParameter.validFile(parameters, "fasta", true);
 156                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 157                         else if (fastafile == "not found") {
 158                                 fastafile = m->getFastaFile();
 159                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 160                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 161                         }else { m->setFastaFile(fastafile); }
 162
 163                         namefile = validParameter.validFile(parameters, "name", true);
 164                         if (namefile == "not open") { namefile = ""; abort = true; }
 165                         else if (namefile == "not found") { namefile = "";  }
 166                         else { m->setNameFile(namefile); }
 167
 168             countfile = validParameter.validFile(parameters, "count", true);
 169                         if (countfile == "not open") { abort = true; countfile = ""; }
 170                         else if (countfile == "not found") { countfile = ""; }
 171                         else { m->setCountTableFile(countfile); }
 172
 173             if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 174
 175                         //if the user changes the output directory command factory will send this info to us in the output parameter
 176                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 177                                 outputDir = "";
 178                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 179                         }
 180
 181             if (countfile == "") {
 182                 if ((namefile == "") && (fastafile != "")){
 183                     vector<string> files; files.push_back(fastafile);
 184                     parser.getNameFile(files);
 185                 }
 186             }
 187                 }
 188
 189         }
 190         catch(exception& e) {
 191                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
 192                 exit(1);
 193         }
 194 }
 195 //**********************************************************************************************************************
 196
 197 int AlignCheckCommand::execute(){
 198         try {
 199
 200                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 201
 202                 //get secondary structure info.
 203                 readMap();
 204
 205                 if (namefile != "") { nameMap = m->readNames(namefile); }
 206         else if (countfile != "") {
 207             CountTable ct;
 208             ct.readTable(countfile);
 209             nameMap = ct.getNameMap();
 210         }
 211
 212                 if (m->control_pressed) { return 0; }
 213
 214                 ifstream in;
 215                 m->openInputFile(fastafile, in);
 216
 217                 ofstream out;
 218                 string outfile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("aligncheck");
 219                 m->openOutputFile(outfile, out);
 220
 221
 222                 out << "name" << '\t' << "pound" << '\t' << "dash" << '\t' << "plus" << '\t' << "equal" << '\t';
 223                 out << "loop" << '\t' << "tilde" << '\t' << "total"  << '\t' << "numseqs" << endl;
 224
 225                 vector<int> pound;
 226                 vector<int> dash;
 227                 vector<int> plus;
 228                 vector<int> equal;
 229                 vector<int> loop;
 230                 vector<int> tilde;
 231                 vector<int> total;
 232
 233                 int count = 0;
 234                 while(!in.eof()){
 235                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outfile); return 0; }
 236
 237                         Sequence seq(in);  m->gobble(in);
 238                         if (seq.getName() != "") {
 239                                 statData data = getStats(seq.getAligned());
 240
 241                                 if (haderror == 1) { m->control_pressed = true; break; }
 242
 243                                 int num = 1;
 244                                 if ((namefile != "") || (countfile != "")) {
 245                                         //make sure this sequence is in the namefile, else error
 246                                         map<string, int>::iterator it = nameMap.find(seq.getName());
 247
 248                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + seq.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 249                                         else { num = it->second; }
 250                                 }
 251
 252                                 //for each sequence this sequence represents
 253                                 for (int i = 0; i < num; i++) {
 254                                         pound.push_back(data.pound);
 255                                         dash.push_back(data.dash);
 256                                         plus.push_back(data.plus);
 257                                         equal.push_back(data.equal);
 258                                         loop.push_back(data.loop);
 259                                         tilde.push_back(data.tilde);
 260                                         total.push_back(data.total);
 261                                 }
 262                                 count++;
 263
 264                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 265                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << '\t' << num << endl;
 266                         }
 267                 }
 268
 269                 in.close();
 270                 out.close();
 271
 272                 if (m->control_pressed) {  m->mothurRemove(outfile); return 0; }
 273
 274                 sort(pound.begin(), pound.end());
 275                 sort(dash.begin(), dash.end());
 276                 sort(plus.begin(), plus.end());
 277                 sort(equal.begin(), equal.end());
 278                 sort(loop.begin(), loop.end());
 279                 sort(tilde.begin(), tilde.end());
 280                 sort(total.begin(), total.end());
 281                 int size = pound.size();
 282
 283                 int ptile0_25   = int(size * 0.025);
 284                 int ptile25             = int(size * 0.250);
 285                 int ptile50             = int(size * 0.500);
 286                 int ptile75             = int(size * 0.750);
 287                 int ptile97_5   = int(size * 0.975);
 288                 int ptile100    = size - 1;
 289
 290                 if (m->control_pressed) {  m->mothurRemove(outfile); return 0; }
 291
 292                 m->mothurOutEndLine();
 293                 m->mothurOut("\t\tPound\tDash\tPlus\tEqual\tLoop\tTilde\tTotal"); m->mothurOutEndLine();
 294                 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();
 295                 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();
 296                 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();
 297                 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();
 298                 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();
 299                 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();
 300                 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();
 301                 if ((namefile == "") && (countfile == "")) {  m->mothurOut("# of Seqs:\t" + toString(count)); m->mothurOutEndLine(); }
 302                 else { m->mothurOut("# of unique seqs:\t" + toString(count)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(size)); m->mothurOutEndLine(); }
 303
 304
 305                 m->mothurOutEndLine();
 306                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 307                 m->mothurOut(outfile); m->mothurOutEndLine();   outputNames.push_back(outfile); outputTypes["aligncheck"].push_back(outfile);
 308                 m->mothurOutEndLine();
 309
 310                 return 0;
 311         }
 312
 313         catch(exception& e) {
 314                 m->errorOut(e, "AlignCheckCommand", "execute");
 315                 exit(1);
 316         }
 317 }
 318 //**********************************************************************************************************************
 319 void AlignCheckCommand::readMap(){
 320         try {
 321
 322                 structMap.resize(1, 0);
 323                 ifstream in;
 324
 325                 m->openInputFile(mapfile, in);
 326
 327                 while(!in.eof()){
 328                         int position;
 329                         in >> position;
 330                         structMap.push_back(position);
 331                         m->gobble(in);
 332                 }
 333                 in.close();
 334
 335                 seqLength = structMap.size();
 336
 337
 338                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 339                 for(int i=0;i<seqLength;i++){
 340                         if(structMap[i] != 0){
 341                                 if(structMap[structMap[i]] != i){
 342                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 343                                 }
 344                         }
 345                 }
 346
 347
 348         }
 349         catch(exception& e) {
 350                 m->errorOut(e, "AlignCheckCommand", "readMap");
 351                 exit(1);
 352         }
 353 }
 354 /**************************************************************************************************/
 355
 356 statData AlignCheckCommand::getStats(string sequence){
 357         try {
 358
 359                 statData data;
 360                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 361
 362                 int length = sequence.length();
 363
 364                 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;  }
 365
 366                 for(int i=1;i<length;i++){
 367                         if(structMap[i] != 0){
 368                                 if(sequence[i] == 'A'){
 369                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 370                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 371                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 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] == 'T'){
 377                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 378                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 379                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 380                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 381                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 382                                         data.total++;
 383                                 }
 384                                 else if(sequence[i] == 'G'){
 385                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 386                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 387                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 388                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 389                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 390                                         data.total++;
 391                                 }
 392                                 else if(sequence[i] == 'C'){
 393                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 394                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 395                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 396                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 397                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 398                                         data.total++;
 399                                 }
 400                                 else if(sequence[i] == '-'){
 401                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 402                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 403                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 404                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 405                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 406                                 }
 407                         }
 408                         else if(isalnum(sequence[i])){
 409                                 data.loop++;
 410                                 data.total++;
 411                         }
 412                 }
 413                 return data;
 414
 415         }
 416         catch(exception& e) {
 417                 m->errorOut(e, "AlignCheckCommand", "getStats");
 418                 exit(1);
 419         }
 420 }
 421 //**********************************************************************************************************************