filterseqscommand.cpp

   1 /*
   2  *  filterseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Thomas Ryabin on 5/4/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "filterseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 /**************************************************************************************/
  14
  15 FilterSeqsCommand::FilterSeqsCommand(string option){
  16         try {
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"fasta", "trump", "soft", "hard", "vertical"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string,string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //check for required parameters
  38                         fastafile = validParameter.validFile(parameters, "fasta", true);
  39                         if (fastafile == "not found") { cout << "fasta is a required parameter for the filter.seqs command." << endl; abort = true; }
  40                         else if (fastafile == "not open") { abort = true; }
  41
  42                         //check for optional parameter and set defaults
  43                         // ...at some point should added some additional type checking...
  44
  45                         string temp;
  46                         temp = validParameter.validFile(parameters, "trump", false);                    if (temp == "not found") { temp = "*"; }
  47                         trump = temp[0];
  48
  49                         temp = validParameter.validFile(parameters, "soft", false);                             if (temp == "not found") { soft = 0; }
  50                         else {  soft = (float)atoi(temp.c_str()) / 100.0;  }
  51
  52                         hard = validParameter.validFile(parameters, "hard", true);                              if (hard == "not found") { hard = ""; }
  53                         else if (hard == "not open") { abort = true; }
  54
  55                         vertical = validParameter.validFile(parameters, "vertical", false);             if (vertical == "not found") { vertical = "T"; }
  56
  57                         numSeqs = 0;
  58
  59                 }
  60
  61         }
  62         catch(exception& e) {
  63                 cout << "Standard Error: " << e.what() << " has occurred in the FilterSeqsCommand class Function FilterSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  64                 exit(1);
  65         }
  66         catch(...) {
  67                 cout << "An unknown error has occurred in the FilterSeqsCommand class function FilterSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  68                 exit(1);
  69         }
  70 }
  71
  72 //**********************************************************************************************************************
  73
  74 void FilterSeqsCommand::help(){
  75         try {
  76                 cout << "The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file." << "\n";
  77                 cout << "The filter.seqs command parameters are fasta, trump, soft, hard and vertical.  " << "\n";
  78                 cout << "The fasta parameter is required." << "\n";
  79                 cout << "The trump parameter .... The default is ..." << "\n";
  80                 cout << "The soft parameter .... The default is ...." << "\n";
  81                 cout << "The hard parameter .... The default is ...." << "\n";
  82                 cout << "The vertical parameter .... The default is T." << "\n";
  83                 cout << "The filter.seqs command should be in the following format: " << "\n";
  84                 cout << "filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) " << "\n";
  85                 cout << "Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T)." << "\n";
  86                 cout << "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta)." << "\n" << "\n";
  87
  88         }
  89         catch(exception& e) {
  90                 cout << "Standard Error: " << e.what() << " has occurred in the FilterSeqsCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  91                 exit(1);
  92         }
  93         catch(...) {
  94                 cout << "An unknown error has occurred in the FilterSeqsCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  95                 exit(1);
  96         }
  97 }
  98
  99 /**************************************************************************************/
 100
 101 void FilterSeqsCommand::doHard() {
 102
 103         ifstream fileHandle;
 104         openInputFile(hard, fileHandle);
 105
 106         fileHandle >> filter;
 107
 108 }
 109
 110 /**************************************************************************************/
 111
 112 void FilterSeqsCommand::doTrump(Sequence seq) {
 113
 114         string curAligned = seq.getAligned();
 115
 116         for(int j = 0; j < alignmentLength; j++) {
 117                 if(curAligned[j] == trump){
 118                         filter[j] = '0';
 119                 }
 120         }
 121
 122 }
 123
 124 /**************************************************************************************/
 125
 126 void FilterSeqsCommand::doVertical() {
 127
 128         for(int i=0;i<alignmentLength;i++){
 129                 if(gap[i] == numSeqs)   {       filter[i] = '0';        }
 130         }
 131
 132 }
 133
 134 /**************************************************************************************/
 135
 136 void FilterSeqsCommand::doSoft() {
 137
 138         int threshold = int (soft * numSeqs);
 139
 140         for(int i=0;i<alignmentLength;i++){
 141                 if(a[i] < threshold && t[i] < threshold && g[i] < threshold && c[i] < threshold){       filter[i] = 0;  }
 142         }
 143 }
 144
 145 /**************************************************************************************/
 146
 147 void FilterSeqsCommand::getFreqs(Sequence seq) {
 148
 149         string curAligned = seq.getAligned();;
 150
 151         for(int j=0;j<alignmentLength;j++){
 152                 if(toupper(curAligned[j]) == 'A')                                                                               {       a[j]++;         }
 153                 else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[j]) == 'U') {       t[j]++;         }
 154                 else if(toupper(curAligned[j]) == 'G')                                                                  {       g[j]++;         }
 155                 else if(toupper(curAligned[j]) == 'C')                                                                  {       c[j]++;         }
 156                 else if(curAligned[j] == '-' || curAligned[j] == '.')                                   {       gap[j]++;       }
 157         }
 158
 159 }
 160
 161 /**************************************************************************************/
 162
 163 int FilterSeqsCommand::execute() {
 164         try {
 165
 166                 if (abort == true) { return 0; }
 167
 168                 ifstream inFASTA;
 169                 openInputFile(fastafile, inFASTA);
 170
 171                 Sequence testSeq(inFASTA);
 172                 alignmentLength = testSeq.getAlignLength();
 173                 inFASTA.seekg(0);
 174
 175                 if(soft != 0 || isTrue(vertical)){
 176                         a.assign(alignmentLength, 0);
 177                         t.assign(alignmentLength, 0);
 178                         g.assign(alignmentLength, 0);
 179                         c.assign(alignmentLength, 0);
 180                         gap.assign(alignmentLength, 0);
 181                 }
 182
 183                 if(hard.compare("") != 0)       {       doHard();               }
 184                 else                                            {       filter = string(alignmentLength, '1');  }
 185
 186                 if(trump != '*' || isTrue(vertical) || soft != 0){
 187                         while(!inFASTA.eof()){  //read through and create the filter...
 188                                 Sequence seq(inFASTA);
 189                                 if(trump != '*'){       doTrump(seq);   }
 190                                 if(isTrue(vertical) || soft != 0){      getFreqs(seq);  }
 191                                 numSeqs++;
 192                                 cout.flush();
 193                         }
 194
 195                 }
 196                 inFASTA.close();
 197
 198                 if(isTrue(vertical) == 1)       {       doVertical();   }
 199                 if(soft != 0)   {       doSoft();               }
 200
 201                 ofstream outFilter;
 202                 string filterFile = getRootName(fastafile) + "filter";
 203                 openOutputFile(filterFile, outFilter);
 204                 outFilter << filter << endl;
 205                 outFilter.close();
 206
 207
 208                 openInputFile(fastafile, inFASTA);
 209                 string filteredFasta = getRootName(fastafile) + "filter.fasta";
 210                 ofstream outFASTA;
 211                 openOutputFile(filteredFasta, outFASTA);
 212
 213                 numSeqs = 0;
 214                 while(!inFASTA.eof()){
 215                         Sequence seq(inFASTA);
 216                         string align = seq.getAligned();
 217                         string filterSeq = "";
 218
 219                         for(int j=0;j<alignmentLength;j++){
 220                                 if(filter[j] == '1'){
 221                                         filterSeq += align[j];
 222                                 }
 223                         }
 224
 225                         outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
 226                         numSeqs++;
 227                         gobble(inFASTA);
 228                 }
 229                 outFASTA.close();
 230                 inFASTA.close();
 231
 232
 233                 int filteredLength = 0;
 234                 for(int i=0;i<alignmentLength;i++){
 235                         if(filter[i] == '1'){   filteredLength++;       }
 236                 }
 237
 238                 cout << endl;
 239                 cout << "Length of filtered alignment: " << filteredLength << endl;
 240                 cout << "Number of columns removed: " << alignmentLength-filteredLength << endl;
 241                 cout << "Length of the original alignment: " << alignmentLength << endl;
 242                 cout << "Number of sequences used to construct filter: " << numSeqs << endl;
 243
 244                 return 0;
 245
 246         }
 247         catch(exception& e) {
 248                 cout << "Standard Error: " << e.what() << " has occurred in the FilterSeqsCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 249                 exit(1);
 250         }
 251         catch(...) {
 252                 cout << "An unknown error has occurred in the FilterSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 253                 exit(1);
 254         }
 255 }
 256
 257 /**************************************************************************************/