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
  15 AlignCheckCommand::AlignCheckCommand(string option)  {
  16         try {
  17                 abort = false;
  18                 haderror = 0;
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string Array[] =  {"fasta","map", "outputdir","inputdir"};
  26                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string,string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string,string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //if the user changes the input directory command factory will send this info to us in the output parameter
  40                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  41                         if (inputDir == "not found"){   inputDir = "";          }
  42                         else {
  43                                 string path;
  44                                 it = parameters.find("fasta");
  45                                 //user has given a template file
  46                                 if(it != parameters.end()){
  47                                         path = m->hasPath(it->second);
  48                                         //if the user has not given a path then, add inputdir. else leave path alone.
  49                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  50                                 }
  51
  52                                 it = parameters.find("map");
  53                                 //user has given a template file
  54                                 if(it != parameters.end()){
  55                                         path = m->hasPath(it->second);
  56                                         //if the user has not given a path then, add inputdir. else leave path alone.
  57                                         if (path == "") {       parameters["map"] = inputDir + it->second;              }
  58                                 }
  59                         }
  60
  61                         //check for required parameters
  62                         mapfile = validParameter.validFile(parameters, "map", true);
  63                         if (mapfile == "not open") { abort = true; }
  64                         else if (mapfile == "not found") {  mapfile = "";  m->mothurOut("You must provide an map file."); m->mothurOutEndLine(); abort = true; }
  65
  66                         fastafile = validParameter.validFile(parameters, "fasta", true);
  67                         if (fastafile == "not open") { abort = true; }
  68                         else if (fastafile == "not found") {  fastafile = "";  m->mothurOut("You must provide an fasta file."); m->mothurOutEndLine(); abort = true;  }
  69
  70                         //if the user changes the output directory command factory will send this info to us in the output parameter
  71                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  72                                 outputDir = "";
  73                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
  74                         }
  75
  76                 }
  77
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "AlignCheckCommand", "RemoveSeqsCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85
  86 void AlignCheckCommand::help(){
  87         try {
  88                 m->mothurOut("The align.check command reads a fasta file and map file.\n");
  89                 m->mothurOut("It outputs a file containing the secondary structure matches in the .align.check file.\n");
  90                 m->mothurOut("The align.check command parameters are fasta and map, both are required.\n");
  91                 m->mothurOut("The align.check command should be in the following format: align.check(fasta=yourFasta, map=yourMap).\n");
  92                 m->mothurOut("Example align.check(map=silva.ss.map, fasta=amazon.fasta).\n");
  93                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  94         }
  95         catch(exception& e) {
  96                 m->errorOut(e, "AlignCheckCommand", "help");
  97                 exit(1);
  98         }
  99 }
 100
 101 //**********************************************************************************************************************
 102
 103 int AlignCheckCommand::execute(){
 104         try {
 105
 106                 if (abort == true) { return 0; }
 107
 108                 //get secondary structure info.
 109                 readMap();
 110
 111                 ifstream in;
 112                 m->openInputFile(fastafile, in);
 113
 114                 ofstream out;
 115                 string outfile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "align.check";
 116                 m->openOutputFile(outfile, out);
 117
 118                 out << "name" << '\t' << "pound" << '\t' << "dash" << '\t' << "plus" << '\t' << "equal" << '\t';
 119                 out << "loop" << '\t' << "tilde" << '\t' << "total" << endl;
 120
 121
 122                 while(!in.eof()){
 123                         if (m->control_pressed) { in.close(); out.close(); remove(outfile.c_str()); return 0; }
 124
 125                         Sequence seq(in);  m->gobble(in);
 126                         if (seq.getName() != "") {
 127                                 statData data = getStats(seq.getAligned());
 128
 129                                 if (haderror == 1) { break; }
 130
 131                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 132                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << endl;
 133                         }
 134                 }
 135
 136                 in.close();
 137                 out.close();
 138
 139                 if (m->control_pressed) {  remove(outfile.c_str()); return 0; }
 140
 141                 m->mothurOutEndLine();
 142                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 143                 m->mothurOut(outfile); m->mothurOutEndLine();
 144                 m->mothurOutEndLine();
 145
 146                 return 0;
 147         }
 148
 149         catch(exception& e) {
 150                 m->errorOut(e, "AlignCheckCommand", "execute");
 151                 exit(1);
 152         }
 153 }
 154 //**********************************************************************************************************************
 155 void AlignCheckCommand::readMap(){
 156         try {
 157
 158                 structMap.resize(1, 0);
 159                 ifstream in;
 160
 161                 m->openInputFile(mapfile, in);
 162
 163                 while(!in.eof()){
 164                         int position;
 165                         in >> position;
 166                         structMap.push_back(position);
 167                         m->gobble(in);
 168                 }
 169                 in.close();
 170
 171                 seqLength = structMap.size();
 172
 173
 174                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 175                 for(int i=0;i<seqLength;i++){
 176                         if(structMap[i] != 0){
 177                                 if(structMap[structMap[i]] != i){
 178                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 179                                 }
 180                         }
 181                 }
 182
 183
 184         }
 185         catch(exception& e) {
 186                 m->errorOut(e, "AlignCheckCommand", "readMap");
 187                 exit(1);
 188         }
 189 }
 190 /**************************************************************************************************/
 191
 192 statData AlignCheckCommand::getStats(string sequence){
 193         try {
 194
 195                 statData data;
 196                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 197
 198                 int length = sequence.length();
 199
 200                 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;  }
 201
 202                 for(int i=1;i<length;i++){
 203                         if(structMap[i] != 0){
 204                                 if(sequence[i] == 'A'){
 205                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 206                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 207                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 208                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 209                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 210                                         data.total++;
 211                                 }
 212                                 else if(sequence[i] == 'T'){
 213                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 214                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 215                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 216                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 217                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 218                                         data.total++;
 219                                 }
 220                                 else if(sequence[i] == 'G'){
 221                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 222                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 223                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 224                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 225                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 226                                         data.total++;
 227                                 }
 228                                 else if(sequence[i] == 'C'){
 229                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 230                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 231                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 232                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 233                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 234                                         data.total++;
 235                                 }
 236                                 else if(sequence[i] == '-'){
 237                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 238                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 239                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 240                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 241                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 242                                 }
 243                         }
 244                         else if(isalnum(sequence[i])){
 245                                 data.loop++;
 246                                 data.total++;
 247                         }
 248                 }
 249                 return data;
 250
 251         }
 252         catch(exception& e) {
 253                 m->errorOut(e, "AlignCheckCommand", "getStats");
 254                 exit(1);
 255         }
 256 }
 257
 258
 259 //**********************************************************************************************************************