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") { mothurOut("fasta is a required parameter for the filter.seqs command."); mothurOutEndLine(); 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                 errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
  64                 exit(1);
  65         }
  66 }
  67
  68 //**********************************************************************************************************************
  69
  70 void FilterSeqsCommand::help(){
  71         try {
  72                 mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
  73                 mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
  74                 mothurOut("The fasta parameter is required.\n");
  75                 mothurOut("The trump parameter .... The default is ...\n");
  76                 mothurOut("The soft parameter .... The default is ....\n");
  77                 mothurOut("The hard parameter .... The default is ....\n");
  78                 mothurOut("The vertical parameter .... The default is T.\n");
  79                 mothurOut("The filter.seqs command should be in the following format: \n");
  80                 mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
  81                 mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
  82                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  83
  84         }
  85         catch(exception& e) {
  86                 errorOut(e, "FilterSeqsCommand", "help");
  87                 exit(1);
  88         }
  89 }
  90
  91 /**************************************************************************************/
  92
  93 void FilterSeqsCommand::doHard() {
  94
  95         ifstream fileHandle;
  96         openInputFile(hard, fileHandle);
  97
  98         fileHandle >> filter;
  99
 100         fileHandle.close();
 101
 102 }
 103
 104 /**************************************************************************************/
 105
 106 void FilterSeqsCommand::doTrump(Sequence seq) {
 107
 108         string curAligned = seq.getAligned();
 109
 110         for(int j = 0; j < alignmentLength; j++) {
 111                 if(curAligned[j] == trump){
 112                         filter[j] = '0';
 113                 }
 114         }
 115
 116 }
 117
 118 /**************************************************************************************/
 119
 120 void FilterSeqsCommand::doVertical() {
 121
 122         for(int i=0;i<alignmentLength;i++){
 123                 if(gap[i] == numSeqs)   {       filter[i] = '0';        }
 124         }
 125
 126 }
 127
 128 /**************************************************************************************/
 129
 130 void FilterSeqsCommand::doSoft() {
 131
 132         int threshold = int (soft * numSeqs);
 133
 134         for(int i=0;i<alignmentLength;i++){
 135                 if(a[i] < threshold && t[i] < threshold && g[i] < threshold && c[i] < threshold){       filter[i] = 0;  }
 136         }
 137 }
 138
 139 /**************************************************************************************/
 140
 141 void FilterSeqsCommand::getFreqs(Sequence seq) {
 142
 143         string curAligned = seq.getAligned();;
 144
 145         for(int j=0;j<alignmentLength;j++){
 146                 if(toupper(curAligned[j]) == 'A')                                                                               {       a[j]++;         }
 147                 else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[j]) == 'U') {       t[j]++;         }
 148                 else if(toupper(curAligned[j]) == 'G')                                                                  {       g[j]++;         }
 149                 else if(toupper(curAligned[j]) == 'C')                                                                  {       c[j]++;         }
 150                 else if(curAligned[j] == '-' || curAligned[j] == '.')                                   {       gap[j]++;       }
 151         }
 152
 153 }
 154
 155 /**************************************************************************************/
 156
 157 int FilterSeqsCommand::execute() {
 158         try {
 159
 160                 if (abort == true) { return 0; }
 161
 162                 ifstream inFASTA;
 163                 openInputFile(fastafile, inFASTA);
 164
 165                 Sequence testSeq(inFASTA);
 166                 alignmentLength = testSeq.getAlignLength();
 167                 inFASTA.seekg(0);
 168
 169                 if(soft != 0 || isTrue(vertical)){
 170                         a.assign(alignmentLength, 0);
 171                         t.assign(alignmentLength, 0);
 172                         g.assign(alignmentLength, 0);
 173                         c.assign(alignmentLength, 0);
 174                         gap.assign(alignmentLength, 0);
 175                 }
 176
 177                 if(hard.compare("") != 0)       {       doHard();               }
 178                 else                                            {       filter = string(alignmentLength, '1');  }
 179
 180                 if(trump != '*' || isTrue(vertical) || soft != 0){
 181                         while(!inFASTA.eof()){  //read through and create the filter...
 182                                 Sequence seq(inFASTA);
 183                                 if(trump != '*'){       doTrump(seq);   }
 184                                 if(isTrue(vertical) || soft != 0){      getFreqs(seq);  }
 185                                 numSeqs++;
 186                                 cout.flush();
 187                         }
 188
 189                 }
 190                 inFASTA.close();
 191
 192                 if(isTrue(vertical) == 1)       {       doVertical();   }
 193                 if(soft != 0)   {       doSoft();               }
 194
 195                 ofstream outFilter;
 196                 string filterFile = getRootName(fastafile) + "filter";
 197                 openOutputFile(filterFile, outFilter);
 198                 outFilter << filter << endl;
 199                 outFilter.close();
 200
 201                 ifstream inFasta2;
 202                 openInputFile(fastafile, inFasta2);
 203                 string filteredFasta = getRootName(fastafile) + "filter.fasta";
 204                 ofstream outFASTA;
 205                 openOutputFile(filteredFasta, outFASTA);
 206
 207                 numSeqs = 0;
 208                 while(!inFasta2.eof()){
 209                         Sequence seq(inFasta2);
 210                         string align = seq.getAligned();
 211                         string filterSeq = "";
 212
 213                         for(int j=0;j<alignmentLength;j++){
 214                                 if(filter[j] == '1'){
 215                                         filterSeq += align[j];
 216                                 }
 217                         }
 218
 219                         outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
 220                         numSeqs++;
 221                         gobble(inFasta2);
 222                 }
 223                 outFASTA.close();
 224                 inFasta2.close();
 225
 226
 227                 int filteredLength = 0;
 228                 for(int i=0;i<alignmentLength;i++){
 229                         if(filter[i] == '1'){   filteredLength++;       }
 230                 }
 231
 232                 mothurOutEndLine();
 233                 mothurOut("Length of filtered alignment: " + toString(filteredLength)); mothurOutEndLine();
 234                 mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); mothurOutEndLine();
 235                 mothurOut("Length of the original alignment: " + toString(alignmentLength)); mothurOutEndLine();
 236                 mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); mothurOutEndLine();
 237
 238                 return 0;
 239
 240         }
 241         catch(exception& e) {
 242                 errorOut(e, "FilterSeqsCommand", "execute");
 243                 exit(1);
 244         }
 245 }
 246
 247 /**************************************************************************************/