X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=filterseqscommand.cpp;h=a6bd54982e9ca509463f5d40822d1ae232e83bb1;hb=5b7ac70116137b52dd7884b76c5bca660a5fea38;hp=aff959ab9359ef0d199ac785d8c43a942ea59255;hpb=c196b6b4768ccb84955d773ff0f22e4994d1ba7b;p=mothur.git

diff --git a/filterseqscommand.cpp b/filterseqscommand.cpp
index aff959a..a6bd549 100644
--- a/filterseqscommand.cpp
+++ b/filterseqscommand.cpp
@@ -3,158 +3,176 @@
  *  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() != "")		{	readSeqs =  new ReadFasta(globaldata->inputFileName);	}
+	else if(globaldata->getNexusFile() != "")	{	readSeqs = new ReadNexus(globaldata->inputFileName);	}
+	else if(globaldata->getClustalFile() != "") {	readSeqs = new ReadClustal(globaldata->inputFileName);	}
+	else if(globaldata->getPhylipFile() != "")	{	readSeqs = new ReadPhylip(globaldata->inputFileName);	}
+	
+	readSeqs->read();
+	db = readSeqs->getDB();
+	numSeqs = db->size();
+	
+	alignmentLength = db->get(0).getAlignLength();
+
+	filter = string(alignmentLength, '1');
 }
 
 /**************************************************************************************/
-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;
-//	}
+
+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::doFilter() {}
+
 /**************************************************************************************/
-int FilterSeqsCommand::execute() {	
-	try {
-		globaldata = GlobalData::getInstance();
-		filename = globaldata->inputFileName;
-		
-		if(globaldata->getFastaFile().compare("") != 0) {
-			readFasta = new ReadFasta(filename);
-			readFasta->read();
-			db = readFasta->getDB();
-		}
-		
-		else if(globaldata->getNexusFile().compare("") != 0) {
-			readNexus = new ReadNexus(filename);
-			readNexus->read();
-			db = readNexus->getDB();
+
+void FilterSeqsCommand::doTrump() {
+
+	char trump = globaldata->getTrump()[0];
+	
+	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';
+			}
 		}
+	}
+
+}
+
+/**************************************************************************************/
+
+void FilterSeqsCommand::doVertical() {
+
+	vector<int> counts(alignmentLength, 0);
+	
+	for(int i = 0; i < numSeqs; i++) {
+		string curAligned = db->get(i).getAligned();;
 		
-		else if(globaldata->getClustalFile().compare("") != 0) {
-			readClustal = new ReadClustal(filename);
-			readClustal->read();
-			db = readClustal->getDB();
+		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';		}
+	}
+}
 
-		else if(globaldata->getPhylipFile().compare("") != 0) {
-			readPhylip = new ReadPhylip(filename);
-			readPhylip->read();
-			db = readPhylip->getDB();
-		}
+/**************************************************************************************/
+
+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 < db->get(0).getLength(); i++) 
-			columnsToRemove[i] = false;
-			
-		// Trump
-		if(globaldata->getTrump().compare("") != 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]++;	}
 		}
-		
-		// Soft
-		if(globaldata->getSoft().compare("") != 0) {}
+	}
 
+	for(int i=0;i<alignmentLength;i++){
+		if(a[i] < softThreshold && t[i] < softThreshold && g[i] < softThreshold && c[i] < softThreshold){
+			filter[i] = '0';			
+		}
+	}
+}
+
+/**************************************************************************************/
+
+int FilterSeqsCommand::execute() {	
+	try {
+						
+		if(globaldata->getHard().compare("") != 0)		{	doHard();		}	//	has to be applied first!
+		if(globaldata->getTrump().compare("") != 0)		{	doTrump();		}
+		if(globaldata->getVertical() == "T")			{	doVertical();	}
+		if(globaldata->getSoft().compare("") != 0)		{	doSoft();		}
+
+		ofstream outfile;
+		string filterFile = getRootName(globaldata->inputFileName) + "filter";
+		openOutputFile(filterFile, outfile);
+
+		outfile << filter << endl;
+		outfile.close();
 		
-			
+		string filteredFasta = getRootName(globaldata->inputFileName) + "filter.fasta";
+		openOutputFile(filteredFasta, outfile);
+
+		for(int i=0;i<numSeqs;i++){
+			string curAligned = db->get(i).getAligned();
+			outfile << '>' << db->get(i).getName() << endl;
+			for(int j=0;j<alignmentLength;j++){
+				if(filter[j] == '1'){
+					outfile << curAligned[j];
+				}
+			}
+			outfile << endl;
+		}
+		outfile.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);
 	}
 }
+
 /**************************************************************************************/