X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=filterseqscommand.cpp;h=68450fb15a7b140803aab2d54d8c5bc84e7ce688;hb=7e354c9abb09ea3cf5b500a16cc7f6dd79ccb6f5;hp=f4cc6d82fbe5d97779d4d80dbba9ee9a483940ab;hpb=37a2c0b7dc68bfa53c8f38a713259929c4c46a9f;p=mothur.git

diff --git a/filterseqscommand.cpp b/filterseqscommand.cpp
index f4cc6d8..68450fb 100644
--- a/filterseqscommand.cpp
+++ b/filterseqscommand.cpp
@@ -8,167 +8,185 @@
  */
 
 #include "filterseqscommand.h"
+#include "sequence.hpp"
 
 /**************************************************************************************/
-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(); j++) {
-			string curChar = curAligned.substr(j, 1);
-			if(curChar.compare(trump) == 0) 
-				columnsToRemove[j] = true;
-		}
-	}
-}
 
-/**************************************************************************************/
-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.push_back(symbols);
-		columnSymbolSums.push_back(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(); j++) {
-			string curChar = curAligned.substr(j, 1);
-			vector<string> curColumnSymbols = columnSymbols[j];
-			bool newSymbol = true;
+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)));
 			
-			for(int k = 0; k < curColumnSymbols.size(); k++) 
-				if(curChar.compare(curColumnSymbols[k]) == 0) {
-					newSymbol = false;
-					columnSymbolSums[j][k]++;
-				}
+			OptionParser parser(option);
+			map<string,string> parameters = parser.getParameters();
+			
+			ValidParameters validParameter;
 			
-			if(newSymbol) {
-				columnSymbols[j].push_back(curChar);
-				columnSymbolSums[j].push_back(1);
+			//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);	}
+			
+			}
+						
 		}
-	}
-	
-	
-	for(int i = 0; i < columnSymbolSums.size(); i++) {
-		int totalSum = 0;
-		int max = 0;
-		vector<int> curColumnSymbols = columnSymbolSums[i];
-		
-		for(int j = 0; j < curColumnSymbols.size(); j++) {
-			int curSum = curColumnSymbols[j];
-			//cout << columnSymbols[i][j] << ": " << curSum << "\n";
-			if(curSum > max)
-				max = curSum;
-			totalSum += curSum;
-		}
-		//cout << "\n";
 		
-		if((double)max/(double)totalSum * 100 < soft)
-			columnsToRemove[i] = true;
+	}
+	catch(exception& e) {
+		errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
+		exit(1);
 	}
 }
 
-/**************************************************************************************/
-void FilterSeqsCommand::doFilter() {
-	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++;
+//**********************************************************************************************************************
+
+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 {
-		globaldata = GlobalData::getInstance();
-		filename = globaldata->inputFileName;
-		
-		if(globaldata->getFastaFile() != "") {
-			readSeqs =  new ReadFasta(filename); }
-		else if(globaldata->getNexusFile() != "") {
-			readSeqs = new ReadNexus(filename); }
-		else if(globaldata->getClustalFile() != "") {
-			readSeqs = new ReadClustal(filename); }
-		else if(globaldata->getPhylipFile() != "") {
-			readSeqs = new ReadPhylip(filename); }
-			
-		readSeqs->read();
-		db = readSeqs->getDB();
+	
+		if (abort == true) { return 0; }
+		
+		ifstream inFASTA;
+		openInputFile(fastafile, inFASTA);
+		
+		Sequence testSeq(inFASTA);
+		alignmentLength = testSeq.getAlignLength();
+		inFASTA.seekg(0);
+		
+		F.setLength(alignmentLength);
 		
-		//for(int i = 0; i < db->size(); i++) {
-//			cout << db->get(i).getLength() << "\n" << db->get(i).getName() << ": " << db->get(i).getAligned() << "\n\n";
-//		}
+		if(soft != 0 || isTrue(vertical)){
+			F.initialize();
+		}
+		
+		if(hard.compare("") != 0)	{	F.doHard(hard);		}
+		else						{	F.setFilter(string(alignmentLength, '1'));	}
 
-		for(int i = 0; i < db->get(0).getLength(); i++) 
-			columnsToRemove.push_back(false);
-		
-				
-		if(globaldata->getTrump().compare("") != 0) 
-			doTrump();
-		else if(globaldata->getSoft().compare("") != 0)
-			doSoft();
-		else if(globaldata->getFilter().compare("") != 0) 
-			doFilter();
-		
-		//for(int i = 0; i < columnsToRemove.size(); i++)
-//		{
-//			cout << "Remove Column " << i << " = ";
-//			if(columnsToRemove[i])
-//				cout << "true\n";
-//			else
-//				cout << "false\n";
-//		}
+		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();
+			}
+		
+		}
+		inFASTA.close();
+		F.setNumSeqs(numSeqs);
+		
+		
+		if(isTrue(vertical) == 1)	{	F.doVertical();	}
+		if(soft != 0)				{	F.doSoft();		}
+		
+		filter = F.getFilter();
 
+		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);
 
-		//Creating the new SequenceDB 
-		SequenceDB newDB;
-		for(int i = 0; i < db->size(); i++) {
-			Sequence curSeq = db->get(i);
-			string curAligned = curSeq.getAligned();
-			string curName = curSeq.getName();
-			string newAligned = "";
-			for(int j = 0; j < curAligned.length(); j++) 
-				if(!columnsToRemove[j]) 
-					newAligned += curAligned.substr(j, 1);
-			
-			Sequence newSeq(curName, newAligned);
-			newDB.add(newSeq);
+		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'){
+					filterSeq += align[j];
+				}
+			}
+
+			outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
+			numSeqs++;
+			gobble(inFasta2);
 		}
+		outFASTA.close();
+		inFasta2.close();
 		
-		string newFileName = getRootName(filename) + "filter.fa";
-		ofstream outfile;
-		outfile.open(newFileName.c_str());
-		newDB.print(outfile);
-		outfile.close();
-			
-		globaldata->clear();
-		//delete db;
-		//delete newDB;
+		
+		int filteredLength = 0;
+		for(int i=0;i<alignmentLength;i++){
+			if(filter[i] == '1'){	filteredLength++;	}
+		}
+		
+		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);
 	}
-}
\ No newline at end of file
+}
+
+/**************************************************************************************/