*/
#include "filterseqscommand.h"
+#include "sequence.hpp"
/**************************************************************************************/
-FilterSeqsCommand::FilterSeqsCommand(){
- globaldata = GlobalData::getInstance();
-
- if(globaldata->getFastaFile() == "") { cout << "You must enter a fasta formatted file" << endl; }
- trump = globaldata->getTrump()[0];
- vertical =
-// readSeqs->read();
-// db = readSeqs->getDB();
-// numSeqs = db->size();
-//
-// alignmentLength = db->get(0).getAlignLength();
-//
-// filter = string(alignmentLength, '1');
-}
-
-/**************************************************************************************/
-
-void FilterSeqsCommand::doHard() {
-
-// string hardName = globaldata->getHard();
-// string hardFilter = "";
-//
-// ifstream fileHandle;
-// openInputFile(hardName, fileHandle);
-//
-// fileHandle >> hardFilter;
-//
-// if(hardFilter.length() != filter.length()){
-// cout << "The hard filter is not the same length as the alignment: Hard filter will not be applied." << endl;
-// }
-// else{
-// filter = hardFilter;
-// }
-
-}
-
-/**************************************************************************************/
-
-void FilterSeqsCommand::doTrump() {
-
-
- for(int i = 0; i < numSeqs; i++) {
- string curAligned = db->get(i).getAligned();;
-
- for(int j = 0; j < alignmentLength; j++) {
- if(curAligned[j] == trump){
- filter[j] = '0';
+FilterSeqsCommand::FilterSeqsCommand(string option){
+ try {
+ abort = false;
+
+ //allow user to run help
+ if(option == "help") { help(); abort = true; }
+
+ else {
+ //valid paramters for this command
+ string Array[] = {"fasta", "trump", "soft", "hard", "vertical"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+
+ OptionParser parser(option);
+ map<string,string> parameters = parser.getParameters();
+
+ ValidParameters validParameter;
+
+ //check to make sure all parameters are valid for command
+ for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
+ if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
}
+
+ //check for required parameters
+ fastafile = validParameter.validFile(parameters, "fasta", true);
+ if (fastafile == "not found") { mothurOut("fasta is a required parameter for the filter.seqs command."); mothurOutEndLine(); abort = true; }
+ else if (fastafile == "not open") { abort = true; }
+
+ //check for optional parameter and set defaults
+ // ...at some point should added some additional type checking...
+
+ string temp;
+ temp = validParameter.validFile(parameters, "trump", false); if (temp == "not found") { temp = "*"; }
+ trump = temp[0];
+
+ temp = validParameter.validFile(parameters, "soft", false); if (temp == "not found") { soft = 0; }
+ else { soft = (float)atoi(temp.c_str()) / 100.0; }
+
+ hard = validParameter.validFile(parameters, "hard", true); if (hard == "not found") { hard = ""; }
+ else if (hard == "not open") { abort = true; }
+
+ vertical = validParameter.validFile(parameters, "vertical", false); if (vertical == "not found") { vertical = "T"; }
+
+ numSeqs = 0;
+
+ if (abort == false) {
+
+ if (soft != 0) { F.setSoft(soft); }
+ if (trump != '*') { F.setTrump(trump); }
+
+ }
+
}
+
+ }
+ catch(exception& e) {
+ errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
+ exit(1);
}
-
-}
-
-/**************************************************************************************/
-
-void FilterSeqsCommand::doVertical() {
-
-// vector<int> counts(alignmentLength, 0);
-//
-// for(int i = 0; i < numSeqs; i++) {
-// string curAligned = db->get(i).getAligned();;
-//
-// for(int j = 0; j < alignmentLength; j++) {
-// if(curAligned[j] == '-' || curAligned[j] == '.'){
-// counts[j]++;
-// }
-// }
-// }
-// for(int i=0;i<alignmentLength;i++){
-// if(counts[i] == numSeqs) { filter[i] = '0'; }
-// }
}
-/**************************************************************************************/
-
-void FilterSeqsCommand::doSoft() {
+//**********************************************************************************************************************
-// int softThreshold = numSeqs * (float)atoi(globaldata->getSoft().c_str()) / 100.0;
-//
-// vector<int> a(alignmentLength, 0);
-// vector<int> t(alignmentLength, 0);
-// vector<int> g(alignmentLength, 0);
-// vector<int> c(alignmentLength, 0);
-// vector<int> x(alignmentLength, 0);
-//
-// for(int i=0;i<numSeqs;i++){
-// string curAligned = db->get(i).getAligned();;
-//
-// for(int j=0;j<alignmentLength;j++){
-// if(toupper(curAligned[j]) == 'A') { a[j]++; }
-// else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[i]) == 'U') { t[j]++; }
-// else if(toupper(curAligned[j]) == 'G') { g[j]++; }
-// else if(toupper(curAligned[j]) == 'C') { c[j]++; }
-// }
-// }
-//
-// for(int i=0;i<alignmentLength;i++){
-// if(a[i] < softThreshold && t[i] < softThreshold && g[i] < softThreshold && c[i] < softThreshold){
-// filter[i] = '0';
-// }
-// }
+void FilterSeqsCommand::help(){
+ try {
+ mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
+ mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
+ mothurOut("The fasta parameter is required.\n");
+ mothurOut("The trump parameter .... The default is ...\n");
+ mothurOut("The soft parameter .... The default is ....\n");
+ mothurOut("The hard parameter .... The default is ....\n");
+ mothurOut("The vertical parameter .... The default is T.\n");
+ mothurOut("The filter.seqs command should be in the following format: \n");
+ mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
+ mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
+ mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
+
+ }
+ catch(exception& e) {
+ errorOut(e, "FilterSeqsCommand", "help");
+ exit(1);
+ }
}
/**************************************************************************************/
int FilterSeqsCommand::execute() {
try {
+
+ if (abort == true) { return 0; }
+
ifstream inFASTA;
- openInputFile(globaldata->getFastaFile(), inFASTA);
+ openInputFile(fastafile, inFASTA);
- Sequence currSequence(inFASTA);
- alignmentLength = currSequence.getAlignLength();
+ Sequence testSeq(inFASTA);
+ alignmentLength = testSeq.getAlignLength();
+ inFASTA.seekg(0);
- //while
+ F.setLength(alignmentLength);
+ if(soft != 0 || isTrue(vertical)){
+ F.initialize();
+ }
- if(globaldata->getHard().compare("") != 0) { doHard(); } // has to be applied first!
- if(globaldata->getTrump().compare("") != 0) { doTrump(); }
- if(isTrue(globaldata->getVertical()) == true) { doVertical(); }
- if(globaldata->getSoft().compare("") != 0) { doSoft(); }
+ if(hard.compare("") != 0) { F.doHard(hard); }
+ else { F.setFilter(string(alignmentLength, '1')); }
+
+ if(trump != '*' || isTrue(vertical) || soft != 0){
+ while(!inFASTA.eof()){ //read through and create the filter...
+ Sequence seq(inFASTA);
+ if(trump != '*'){ F.doTrump(seq); }
+ if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
+ numSeqs++;
+ cout.flush();
+ }
- ofstream outfile;
- string filterFile = getRootName(globaldata->inputFileName) + "filter";
- openOutputFile(filterFile, outfile);
-
- outfile << filter << endl;
- outfile.close();
+ }
+ inFASTA.close();
+ F.setNumSeqs(numSeqs);
- string filteredFasta = getRootName(globaldata->inputFileName) + "filter.fasta";
- openOutputFile(filteredFasta, outfile);
+
+ if(isTrue(vertical) == 1) { F.doVertical(); }
+ if(soft != 0) { F.doSoft(); }
+
+ filter = F.getFilter();
- for(int i=0;i<numSeqs;i++){
- string curAligned = db->get(i).getAligned();
- outfile << '>' << db->get(i).getName() << endl;
+ ofstream outFilter;
+ string filterFile = getRootName(fastafile) + "filter";
+ openOutputFile(filterFile, outFilter);
+ outFilter << filter << endl;
+ outFilter.close();
+
+ ifstream inFasta2;
+ openInputFile(fastafile, inFasta2);
+ string filteredFasta = getRootName(fastafile) + "filter.fasta";
+ ofstream outFASTA;
+ openOutputFile(filteredFasta, outFASTA);
+
+ numSeqs = 0;
+ while(!inFasta2.eof()){
+ Sequence seq(inFasta2);
+ string align = seq.getAligned();
+ string filterSeq = "";
+
for(int j=0;j<alignmentLength;j++){
if(filter[j] == '1'){
- outfile << curAligned[j];
+ filterSeq += align[j];
}
}
- outfile << endl;
+
+ outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
+ numSeqs++;
+ gobble(inFasta2);
}
- outfile.close();
+ outFASTA.close();
+ inFasta2.close();
+
int filteredLength = 0;
for(int i=0;i<alignmentLength;i++){
if(filter[i] == '1'){ filteredLength++; }
}
- cout << endl;
- cout << "Length of filtered alignment: " << filteredLength << endl;
- cout << "Number of columns removed: " << alignmentLength-filteredLength << endl;
- cout << "Length of the original alignment: " << alignmentLength << endl;
- cout << "Number of sequences used to construct filter: " << numSeqs << endl;
-
- globaldata->clear();
+ mothurOutEndLine();
+ mothurOut("Length of filtered alignment: " + toString(filteredLength)); mothurOutEndLine();
+ mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); mothurOutEndLine();
+ mothurOut("Length of the original alignment: " + toString(alignmentLength)); mothurOutEndLine();
+ mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); mothurOutEndLine();
return 0;
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the FilterSeqsCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the FilterSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ errorOut(e, "FilterSeqsCommand", "execute");
exit(1);
}
}
projects / mothur.git / blobdiff