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 = 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 = 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 += 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                 openInputFile(fastafile, in);
 113
 114                 ofstream out;
 115                 string outfile = outputDir + getRootName(getSimpleName(fastafile)) + "align.check";
 116                 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
 124                         Sequence seq(in);  gobble(in);
 125                         if (seq.getName() != "") {
 126                                 statData data = getStats(seq.getAligned());
 127
 128                                 if (haderror == 1) { break; }
 129
 130                                 out << seq.getName() << '\t' << data.pound << '\t' << data.dash << '\t' << data.plus << '\t' << data.equal << '\t';
 131                                 out << data.loop << '\t' << data.tilde << '\t' << data.total << endl;
 132                         }
 133                 }
 134
 135                 in.close();
 136                 out.close();
 137
 138                 m->mothurOutEndLine();
 139                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 140                 m->mothurOut(outfile); m->mothurOutEndLine();
 141                 m->mothurOutEndLine();
 142
 143                 return 0;
 144         }
 145
 146         catch(exception& e) {
 147                 m->errorOut(e, "AlignCheckCommand", "execute");
 148                 exit(1);
 149         }
 150 }
 151 //**********************************************************************************************************************
 152 void AlignCheckCommand::readMap(){
 153         try {
 154
 155                 structMap.resize(1, 0);
 156                 ifstream in;
 157
 158                 openInputFile(mapfile, in);
 159
 160                 while(!in.eof()){
 161                         int position;
 162                         in >> position;
 163                         structMap.push_back(position);
 164                         gobble(in);
 165                 }
 166                 in.close();
 167
 168                 seqLength = structMap.size();
 169
 170
 171                 //check you make sure is structMap[10] = 380 then structMap[380] = 10.
 172                 for(int i=0;i<seqLength;i++){
 173                         if(structMap[i] != 0){
 174                                 if(structMap[structMap[i]] != i){
 175                                         m->mothurOut("Your map file contains an error:  line " + toString(i) + " does not match line " + toString(structMap[i]) + "."); m->mothurOutEndLine();
 176                                 }
 177                         }
 178                 }
 179
 180
 181         }
 182         catch(exception& e) {
 183                 m->errorOut(e, "AlignCheckCommand", "readMap");
 184                 exit(1);
 185         }
 186 }
 187 /**************************************************************************************************/
 188
 189 statData AlignCheckCommand::getStats(string sequence){
 190         try {
 191
 192                 statData data;
 193                 sequence = "*" + sequence; // need to pad the sequence so we can index it by 1
 194
 195                 int length = sequence.length();
 196
 197                 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;  }
 198
 199                 for(int i=1;i<length;i++){
 200                         if(structMap[i] != 0){
 201                                 if(sequence[i] == 'A'){
 202                                         if(sequence[structMap[i]] == 'T')               {       data.tilde++;   }
 203                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 204                                         else if(sequence[structMap[i]] == 'G')  {       data.equal++;   }
 205                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 206                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 207                                         data.total++;
 208                                 }
 209                                 else if(sequence[i] == 'T'){
 210                                         if(sequence[structMap[i]] == 'T')               {       data.plus++;    }
 211                                         else if(sequence[structMap[i]] == 'A')  {       data.tilde++;   }
 212                                         else if(sequence[structMap[i]] == 'G')  {       data.dash++;    }
 213                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 214                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 215                                         data.total++;
 216                                 }
 217                                 else if(sequence[i] == 'G'){
 218                                         if(sequence[structMap[i]] == 'T')               {       data.dash++;    }
 219                                         else if(sequence[structMap[i]] == 'A')  {       data.equal++;   }
 220                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   }
 221                                         else if(sequence[structMap[i]] == 'C')  {       data.tilde++;   }
 222                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 223                                         data.total++;
 224                                 }
 225                                 else if(sequence[i] == 'C'){
 226                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   }
 227                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   }
 228                                         else if(sequence[structMap[i]] == 'G')  {       data.tilde++;   }
 229                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   }
 230                                         else if(sequence[structMap[i]] == '-')  {       data.pound++;   }
 231                                         data.total++;
 232                                 }
 233                                 else if(sequence[i] == '-'){
 234                                         if(sequence[structMap[i]] == 'T')               {       data.pound++;   data.total++;   }
 235                                         else if(sequence[structMap[i]] == 'A')  {       data.pound++;   data.total++;   }
 236                                         else if(sequence[structMap[i]] == 'G')  {       data.pound++;   data.total++;   }
 237                                         else if(sequence[structMap[i]] == 'C')  {       data.pound++;   data.total++;   }
 238                                         else if(sequence[structMap[i]] == '-')  {               /*donothing*/                           }
 239                                 }
 240                         }
 241                         else if(isalnum(sequence[i])){
 242                                 data.loop++;
 243                                 data.total++;
 244                         }
 245                 }
 246                 return data;
 247
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "AlignCheckCommand", "getStats");
 251                 exit(1);
 252         }
 253 }
 254
 255
 256 //**********************************************************************************************************************