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","aligncheck",false,true,true); parameters.push_back(pfasta);
  18                 CommandParameter pmap("map", "InputTypes", "", "", "none", "none", "none","",false,true,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::getOutputPattern(string type) {
  52     try {
  53         string pattern = "";
  54
  55         if (type == "aligncheck") {  pattern = "[filename],align.check"; }
  56         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  57
  58         return pattern;
  59     }
  60     catch(exception& e) {
  61         m->errorOut(e, "AlignCheckCommand", "getOutputPattern");
  62         exit(1);
  63     }
  64 }
  65 //**********************************************************************************************************************
  66 AlignCheckCommand::AlignCheckCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["aligncheck"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79
  80 AlignCheckCommand::AlignCheckCommand(string option)  {
  81         try {
  82                 abort = false; calledHelp = false;
  83                 haderror = 0;
  84
  85                 //allow user to run help
  86                 if(option == "help") { help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string,string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string,string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["aligncheck"] = tempOutNames;
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 string path;
 112                                 it = parameters.find("fasta");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 118                                 }
 119
 120                                 it = parameters.find("map");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["map"] = inputDir + it->second;              }
 126                                 }
 127
 128                                 it = parameters.find("name");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 134                                 }
 135
 136                 it = parameters.find("count");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 142                                 }
 143                         }
 144
 145                         //check for required parameters
 146                         mapfile = validParameter.validFile(parameters, "map", true);
 147                         if (mapfile == "not open") { abort = true; }
 148                         else if (mapfile == "not found") {  mapfile = "";  m->mothurOut("You must provide an map file."); m->mothurOutEndLine(); abort = true; }
 149
 150                         fastafile = validParameter.validFile(parameters, "fasta", true);
 151                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 152                         else if (fastafile == "not found") {
 153                                 fastafile = m->getFastaFile();
 154                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 155                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 156                         }else { m->setFastaFile(fastafile); }
 157
 158                         namefile = validParameter.validFile(parameters, "name", true);
 159                         if (namefile == "not open") { namefile = ""; abort = true; }
 160                         else if (namefile == "not found") { namefile = "";  }
 161                         else { m->setNameFile(namefile); }
 162
 163             countfile = validParameter.validFile(parameters, "count", true);
 164                         if (countfile == "not open") { abort = true; countfile = ""; }
 165                         else if (countfile == "not found") { countfile = ""; }
 166                         else { m->setCountTableFile(countfile); }
 167
 168             if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 169
 170                         //if the user changes the output directory command factory will send this info to us in the output parameter
 171                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 172                                 outputDir = "";
 173                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 174                         }
 175
 176             if (countfile == "") {
 177                 if ((namefile == "") && (fastafile != "")){
 178                     vector<string> files; files.push_back(fastafile);
 179                     parser.getNameFile(files);
 180                 }
 181             }
 182                 }
 183
 184         }
 185         catch(exception& e) {
 186                 m->errorOut(e, "AlignCheckCommand", "AlignCheckCommand");
 187                 exit(1);
 188         }
 189 }
 190 //**********************************************************************************************************************
 191
 192 int AlignCheckCommand::execute(){
 193         try {
 194
 195                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 196
 197                 //get secondary structure info.
 198                 readMap();
 199
 200                 if (namefile != "") { nameMap = m->readNames(namefile); }
 201         else if (countfile != "") {
 202             CountTable ct;
 203             ct.readTable(countfile, false);
 204             nameMap = ct.getNameMap();
 205         }
 206
 207                 if (m->control_pressed) { return 0; }
 208
 209                 ifstream in;
 210                 m->openInputFile(fastafile, in);
 211
 212                 ofstream out;
 213         map<string, string> variables;
 214                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 215                 string outfile = getOutputFileName("aligncheck",variables);
 216                 m->openOutputFile(outfile, out);
 217
 218
 219                 out << "name" << '\t' << "pound" << '\t' << "dash" << '\t' << "plus" << '\t' << "equal" << '\t';
 220                 out << "loop" << '\t' << "tilde" << '\t' << "total"  << '\t' << "numseqs" << endl;
 221
 222                 vector<int> pound;
 223                 vector<int> dash;
 224                 vector<int> plus;
 225                 vector<int> equal;
 226                 vector<int> loop;
 227                 vector<int> tilde;
 228                 vector<int> total;
 229
 230                 int count = 0;
 231                 while(!in.eof()){
 232                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outfile); return 0; }
 233
 234                         Sequence seq(in);  m->gobble(in);
 235                         if (seq.getName() != "") {
 236                                 statData data = getStats(seq.getAligned());
 237
 238                                 if (haderror == 1) { m->control_pressed = true; break; }
 239
 240                                 int num = 1;
 241                                 if ((namefile != "") || (countfile != "")) {
 242                                         //make sure this sequence is in the namefile, else error
 243                                         map<string, int>::iterator it = nameMap.find(seq.getName());
 244
 245                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + seq.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 246                                         else { num = it->second; }
 247                                 }
 248
 249                                 //for each sequence this sequence represents
 250                                 for (int i = 0; i < num; i++) {
 251                                         pound.push_back(data.pound);
 252                                         dash.push_back(data.dash);
 253                                         plus.push_back(data.plus);
 254                                         equal.push_back(data.equal);
 255                                         loop.push_back(data.loop);
 256                                         tilde.push_back(data.tilde);
 257                                         total.push_back(data.total);
 258                                 }
 259                                 count++;
 260
 261                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 262                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << '\t' << num << endl;
 263                         }
 264                 }
 265
 266                 in.close();
 267                 out.close();
 268
 269                 if (m->control_pressed) {  m->mothurRemove(outfile); return 0; }
 270
 271                 sort(pound.begin(), pound.end());
 272                 sort(dash.begin(), dash.end());
 273                 sort(plus.begin(), plus.end());
 274                 sort(equal.begin(), equal.end());
 275                 sort(loop.begin(), loop.end());
 276                 sort(tilde.begin(), tilde.end());
 277                 sort(total.begin(), total.end());
 278                 int size = pound.size();
 279
 280                 int ptile0_25   = int(size * 0.025);
 281                 int ptile25             = int(size * 0.250);
 282                 int ptile50             = int(size * 0.500);
 283                 int ptile75             = int(size * 0.750);
 284                 int ptile97_5   = int(size * 0.975);
 285                 int ptile100    = size - 1;
 286
 287                 if (m->control_pressed) {  m->mothurRemove(outfile); return 0; }
 288
 289                 m->mothurOutEndLine();
 290                 m->mothurOut("\t\tPound\tDash\tPlus\tEqual\tLoop\tTilde\tTotal"); m->mothurOutEndLine();
 291                 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();
 292                 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();
 293                 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();
 294                 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();
 295                 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();
 296                 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();
 297                 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();
 298                 if ((namefile == "") && (countfile == "")) {  m->mothurOut("# of Seqs:\t" + toString(count)); m->mothurOutEndLine(); }
 299                 else { m->mothurOut("# of unique seqs:\t" + toString(count)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(size)); m->mothurOutEndLine(); }
 300
 301
 302                 m->mothurOutEndLine();
 303                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 304                 m->mothurOut(outfile); m->mothurOutEndLine();   outputNames.push_back(outfile); outputTypes["aligncheck"].push_back(outfile);
 305                 m->mothurOutEndLine();
 306
 307                 return 0;
 308         }
 309
 310         catch(exception& e) {
 311                 m->errorOut(e, "AlignCheckCommand", "execute");
 312                 exit(1);
 313         }
 314 }
 315 //**********************************************************************************************************************
 316 void AlignCheckCommand::readMap(){
 317         try {
 318
 319                 structMap.resize(1, 0);
 320                 ifstream in;
 321
 322                 m->openInputFile(mapfile, in);
 323
 324                 while(!in.eof()){
 325                         int position;
 326                         in >> position;
 327                         structMap.push_back(position);
 328                         m->gobble(in);
 329                 }
 330                 in.close();
 331
 332                 seqLength = structMap.size();
 333
 334
 335                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 336                 for(int i=0;i<seqLength;i++){
 337                         if(structMap[i] != 0){
 338                                 if(structMap[structMap[i]] != i){
 339                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 340                                 }
 341                         }
 342                 }
 343
 344
 345         }
 346         catch(exception& e) {
 347                 m->errorOut(e, "AlignCheckCommand", "readMap");
 348                 exit(1);
 349         }
 350 }
 351 /**************************************************************************************************/
 352
 353 statData AlignCheckCommand::getStats(string sequence){
 354         try {
 355
 356                 statData data;
 357                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 358
 359                 int length = sequence.length();
 360
 361                 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;  }
 362
 363                 for(int i=1;i<length;i++){
 364                         if(structMap[i] != 0){
 365                                 if(sequence[i] == 'A'){
 366                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 367                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 368                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 369                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 370                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 371                                         data.total++;
 372                                 }
 373                                 else if(sequence[i] == 'T'){
 374                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 375                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 376                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 377                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 378                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 379                                         data.total++;
 380                                 }
 381                                 else if(sequence[i] == 'G'){
 382                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 383                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 384                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 385                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 386                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 387                                         data.total++;
 388                                 }
 389                                 else if(sequence[i] == 'C'){
 390                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 391                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 392                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 393                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 394                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 395                                         data.total++;
 396                                 }
 397                                 else if(sequence[i] == '-'){
 398                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 399                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 400                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 401                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 402                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 403                                 }
 404                         }
 405                         else if(isalnum(sequence[i])){
 406                                 data.loop++;
 407                                 data.total++;
 408                         }
 409                 }
 410                 return data;
 411
 412         }
 413         catch(exception& e) {
 414                 m->errorOut(e, "AlignCheckCommand", "getStats");
 415                 exit(1);
 416         }
 417 }
 418 //**********************************************************************************************************************