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         fileHandle.close();
 109
 110 }
 111
 112 /**************************************************************************************/
 113
 114 void FilterSeqsCommand::doTrump(Sequence seq) {
 115
 116         string curAligned = seq.getAligned();
 117
 118         for(int j = 0; j < alignmentLength; j++) {
 119                 if(curAligned[j] == trump){
 120                         filter[j] = '0';
 121                 }
 122         }
 123
 124 }
 125
 126 /**************************************************************************************/
 127
 128 void FilterSeqsCommand::doVertical() {
 129
 130         for(int i=0;i<alignmentLength;i++){
 131                 if(gap[i] == numSeqs)   {       filter[i] = '0';        }
 132         }
 133
 134 }
 135
 136 /**************************************************************************************/
 137
 138 void FilterSeqsCommand::doSoft() {
 139
 140         int threshold = int (soft * numSeqs);
 141
 142         for(int i=0;i<alignmentLength;i++){
 143                 if(a[i] < threshold && t[i] < threshold && g[i] < threshold && c[i] < threshold){       filter[i] = 0;  }
 144         }
 145 }
 146
 147 /**************************************************************************************/
 148
 149 void FilterSeqsCommand::getFreqs(Sequence seq) {
 150
 151         string curAligned = seq.getAligned();;
 152
 153         for(int j=0;j<alignmentLength;j++){
 154                 if(toupper(curAligned[j]) == 'A')                                                                               {       a[j]++;         }
 155                 else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[j]) == 'U') {       t[j]++;         }
 156                 else if(toupper(curAligned[j]) == 'G')                                                                  {       g[j]++;         }
 157                 else if(toupper(curAligned[j]) == 'C')                                                                  {       c[j]++;         }
 158                 else if(curAligned[j] == '-' || curAligned[j] == '.')                                   {       gap[j]++;       }
 159         }
 160
 161 }
 162
 163 /**************************************************************************************/
 164
 165 int FilterSeqsCommand::execute() {
 166         try {
 167
 168                 if (abort == true) { return 0; }
 169
 170                 ifstream inFASTA;
 171                 openInputFile(fastafile, inFASTA);
 172
 173                 Sequence testSeq(inFASTA);
 174                 alignmentLength = testSeq.getAlignLength();
 175                 inFASTA.seekg(0);
 176
 177                 if(soft != 0 || isTrue(vertical)){
 178                         a.assign(alignmentLength, 0);
 179                         t.assign(alignmentLength, 0);
 180                         g.assign(alignmentLength, 0);
 181                         c.assign(alignmentLength, 0);
 182                         gap.assign(alignmentLength, 0);
 183                 }
 184
 185                 if(hard.compare("") != 0)       {       doHard();               }
 186                 else                                            {       filter = string(alignmentLength, '1');  }
 187
 188                 if(trump != '*' || isTrue(vertical) || soft != 0){
 189                         while(!inFASTA.eof()){  //read through and create the filter...
 190                                 Sequence seq(inFASTA);
 191                                 if(trump != '*'){       doTrump(seq);   }
 192                                 if(isTrue(vertical) || soft != 0){      getFreqs(seq);  }
 193                                 numSeqs++;
 194                                 cout.flush();
 195                         }
 196
 197                 }
 198                 inFASTA.close();
 199
 200                 if(isTrue(vertical) == 1)       {       doVertical();   }
 201                 if(soft != 0)   {       doSoft();               }
 202
 203                 ofstream outFilter;
 204                 string filterFile = getRootName(fastafile) + "filter";
 205                 openOutputFile(filterFile, outFilter);
 206                 outFilter << filter << endl;
 207                 outFilter.close();
 208
 209
 210                 openInputFile(fastafile, inFASTA);
 211                 string filteredFasta = getRootName(fastafile) + "filter.fasta";
 212                 ofstream outFASTA;
 213                 openOutputFile(filteredFasta, outFASTA);
 214
 215                 numSeqs = 0;
 216                 while(!inFASTA.eof()){
 217                         Sequence seq(inFASTA);
 218                         string align = seq.getAligned();
 219                         string filterSeq = "";
 220
 221                         for(int j=0;j<alignmentLength;j++){
 222                                 if(filter[j] == '1'){
 223                                         filterSeq += align[j];
 224                                 }
 225                         }
 226
 227                         outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
 228                         numSeqs++;
 229                         gobble(inFASTA);
 230                 }
 231                 outFASTA.close();
 232                 inFASTA.close();
 233
 234
 235                 int filteredLength = 0;
 236                 for(int i=0;i<alignmentLength;i++){
 237                         if(filter[i] == '1'){   filteredLength++;       }
 238                 }
 239
 240                 cout << endl;
 241                 cout << "Length of filtered alignment: " << filteredLength << endl;
 242                 cout << "Number of columns removed: " << alignmentLength-filteredLength << endl;
 243                 cout << "Length of the original alignment: " << alignmentLength << endl;
 244                 cout << "Number of sequences used to construct filter: " << numSeqs << endl;
 245
 246                 return 0;
 247
 248         }
 249         catch(exception& e) {
 250                 cout << "Standard Error: " << e.what() << " has occurred in the FilterSeqsCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 251                 exit(1);
 252         }
 253         catch(...) {
 254                 cout << "An unknown error has occurred in the FilterSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 255                 exit(1);
 256         }
 257 }
 258
 259 /**************************************************************************************/