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