* Mothur
*
* Created by Thomas Ryabin on 5/4/09.
- * Copyright 2009 __MyCompanyName__. All rights reserved.
+ * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
*
*/
#include "filterseqscommand.h"
-#include <iostream>
-#include <fstream>
/**************************************************************************************/
-void FilterSeqsCommand::doTrump() {
- //trump = globaldata->getTrump();
-//
-// for(int i = 0; i < db->size(); i++) {
-// Sequence cur = db->get(i);
-// string curAligned = cur.getAligned();
-//
-// for(int j = 0; j < curAligned.length-1; j++) {
-// string curChar = curAligned.substr(j, j+1);
-//
-// if(curChar.compare(trump) == 0)
-// columnsToRemove[j] = true;
-// }
-// }
+
+FilterSeqsCommand::FilterSeqsCommand(){
+
+ globaldata = GlobalData::getInstance();
+
+ if(globaldata->getFastaFile() == "") { cout << "You must enter a fasta formatted file" << endl; }
+ trump = globaldata->getTrump()[0];
+ numSeqs = 0;
+
}
/**************************************************************************************/
+
+void FilterSeqsCommand::doHard() {
+
+ string hardName = globaldata->getHard();
+ string hardFilter = "";
+
+ ifstream fileHandle;
+ openInputFile(hardName, fileHandle);
+
+ fileHandle >> filter;
+
+}
+
+/**************************************************************************************/
+
+void FilterSeqsCommand::doTrump(Sequence seq) {
+
+ string curAligned = seq.getAligned();
+
+ for(int j = 0; j < alignmentLength; j++) {
+ if(curAligned[j] == trump){
+ filter[j] = '0';
+ }
+ }
+
+}
+
+/**************************************************************************************/
+
+void FilterSeqsCommand::doVertical() {
+
+ for(int i=0;i<alignmentLength;i++){
+ if(gap[i] == numSeqs) { filter[i] = '0'; }
+ }
+
+}
+
+/**************************************************************************************/
+
void FilterSeqsCommand::doSoft() {
- //soft = atoi(globaldata->getSoft().c_str());
-// vector<vector<int> > columnSymbolSums;
-// vector<vector<string> > columnSymbols;
-// for(int i = 0; i < db->get(0).getLength(); i++) {
-// vector<string> symbols;
-// vector<int> sums;
-// columnSymbols[i] = symbols;
-// columnSymbolSums[i] = sums;
-// }
-//
-// for(int i = 0; i < db->size(); i++) {
-// Sequence cur = db->get(i);
-// string curAligned = cur.getAligned();
-//
-// for(int j = 0; j < curAligned.length-1; j++) {
-// string curChar = curAligned.substr(j, j+1);
-// vector<string> curColumnSymbols = columnSymbols[j];
-//
-// bool newSymbol = true;
-//
-// for(int k = 0; j < curColumnSymbols.size(); j++)
-// if(curChar.compare(curColumnSymbols[k]) == 0) {
-// newSymbol = false;
-// columnSymbolSums[j][k]++;
-// }
-//
-// if(newSymbol) {
-// columnSymbols.push_back(curChar);
-// columnSymbolSums[j].push_back(1);
-// }
-// }
-// }
-//
-// for(int i = 0; i < columnSymbolSums.size(); i++) {
-// int totalSum = 0;
-// int max = 0;
-// vector<int> curColumn = columnSymbolSums[i];
-//
-// for(int j = 0; j < curColumn.size(); j++) {
-// int curSum = curColumn[j];
-// if(curSum > max)
-// max = curSum;
-// totalSum += curSum;
-// }
-//
-// if((double)max/(double)totalSum * 100 < soft)
-// columnsToRemove[i] = true;
-// }
+
+ int threshold = int (soft * numSeqs);
+ bool keep = 0;
+
+ for(int i=0;i<alignmentLength;i++){
+ if(a[i] >= threshold) { keep = 1; }
+ else if(t[i] >= threshold) { keep = 1; }
+ else if(g[i] >= threshold) { keep = 1; }
+ else if(c[i] >= threshold) { keep = 1; }
+
+ if(keep == 0) { filter[i] = 0; }
+ }
}
-void FilterSeqsCommand::doFilter() {}
+
/**************************************************************************************/
+
+void FilterSeqsCommand::getFreqs(Sequence seq) {
+
+ string curAligned = seq.getAligned();;
+
+ for(int j=0;j<alignmentLength;j++){
+ if(toupper(curAligned[j]) == 'A') { a[j]++; }
+ else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[j]) == 'U') { t[j]++; }
+ else if(toupper(curAligned[j]) == 'G') { g[j]++; }
+ else if(toupper(curAligned[j]) == 'C') { c[j]++; }
+ else if(curAligned[j] == '-' || curAligned[j] == '.') { gap[j]++; }
+ }
+
+}
+
+/**************************************************************************************/
+
int FilterSeqsCommand::execute() {
try {
- globaldata = GlobalData::getInstance();
- filename = globaldata->inputFileName;
+ ifstream inFASTA;
+ openInputFile(globaldata->getFastaFile(), inFASTA);
- if(globaldata->getFastaFile().compare("") != 0) {
- readFasta = new ReadFasta(filename);
- readFasta->read();
- db = readFasta->getDB();
- }
+ Sequence testSeq(inFASTA);
+ alignmentLength = testSeq.getAlignLength();
+ inFASTA.seekg(0);
- else if(globaldata->getNexusFile().compare("") != 0) {
- readNexus = new ReadNexus(filename);
- readNexus->read();
- db = readNexus->getDB();
+ if(globaldata->getSoft() != "" || isTrue(globaldata->getVertical())){
+ a.assign(alignmentLength, 0);
+ t.assign(alignmentLength, 0);
+ g.assign(alignmentLength, 0);
+ c.assign(alignmentLength, 0);
+ gap.assign(alignmentLength, 0);
}
-
- else if(globaldata->getClustalFile().compare("") != 0) {
- readClustal = new ReadClustal(filename);
- readClustal->read();
- db = readClustal->getDB();
+ if(globaldata->getSoft() != ""){
+ soft = (float)atoi(globaldata->getSoft().c_str()) / 100.0;
}
+
+ if(globaldata->getHard().compare("") != 0) { doHard(); }
+ else { filter = string(alignmentLength, '1'); }
- else if(globaldata->getPhylipFile().compare("") != 0) {
- readPhylip = new ReadPhylip(filename);
- readPhylip->read();
- db = readPhylip->getDB();
- }
-
- for(int i = 0; i < db->get(0).getLength(); i++)
- columnsToRemove[i] = false;
-
- // Trump
- if(globaldata->getTrump().compare("") != 0) {
+ if(globaldata->getTrump().compare("") != 0 || isTrue(globaldata->getVertical()) || globaldata->getSoft().compare("") != 0){
+
+ while(!inFASTA.eof()){
+ Sequence seq(inFASTA);
+ if(globaldata->getTrump().compare("") != 0) { doTrump(seq); }
+ if(isTrue(globaldata->getVertical()) || globaldata->getSoft().compare("") != 0){ getFreqs(seq); }
+ numSeqs++;
+ cout.flush();
+ }
-
}
+ inFASTA.close();
+
+ if(isTrue(globaldata->getVertical()) == 1) { doVertical(); }
+ if(globaldata->getSoft().compare("") != 0) { doSoft(); }
+
+ ofstream outFilter;
+ string filterFile = getRootName(globaldata->inputFileName) + "filter";
+ openOutputFile(filterFile, outFilter);
+ outFilter << filter << endl;
+ outFilter.close();
- // Soft
- if(globaldata->getSoft().compare("") != 0) {}
+ openInputFile(globaldata->getFastaFile(), inFASTA);
+ string filteredFasta = getRootName(globaldata->inputFileName) + "filter.fasta";
+ ofstream outFASTA;
+ openOutputFile(filteredFasta, outFASTA);
+
+ numSeqs = 0;
+ while(!inFASTA.eof()){
+ Sequence seq(inFASTA);
+ string align = seq.getAligned();
+ string filterSeq = "";
+
+ for(int j=0;j<alignmentLength;j++){
+ if(filter[j] == '1'){
+ filterSeq += align[j];
+ }
+ }
+
+ outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
+ numSeqs++;
+ gobble(inFASTA);
+ }
+ outFASTA.close();
+ inFASTA.close();
-
- // Filter
- //if(globaldata->getFilter().compare("") != 0) {
-//
-// filter = globaldata->getFilter();
-// ifstream filehandle;
-// openInputFile(filter, filehandle);
-//
-// char c;
-// int count = 0;
-// while(!filehandle.eof()) {
-// c = filehandle.get();
-// if(c == '0')
-// columnsToRemove[count] = true;
-// count++;
-// }
-// }
+ 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();
-
-
return 0;
+
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the DeconvoluteCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ 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 DeconvoluteCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ cout << "An unknown error has occurred in the FilterSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
exit(1);
}
}
+
/**************************************************************************************/
projects / mothur.git / blobdiff