*/
#include "deconvolutecommand.h"
+#include "sequence.hpp"
//**********************************************************************************************************************
-vector<string> DeconvoluteCommand::getValidParameters(){
+vector<string> DeconvoluteCommand::setParameters(){
try {
- string Array[] = {"fasta", "name","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
return myArray;
}
catch(exception& e) {
- m->errorOut(e, "DeconvoluteCommand", "getValidParameters");
+ m->errorOut(e, "DeconvoluteCommand", "setParameters");
exit(1);
}
}
//**********************************************************************************************************************
-DeconvoluteCommand::DeconvoluteCommand(){
+string DeconvoluteCommand::getHelpString(){
try {
- abort = true;
- //initialize outputTypes
- vector<string> tempOutNames;
- outputTypes["fasta"] = tempOutNames;
- outputTypes["name"] = tempOutNames;
- }
- catch(exception& e) {
- m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
- exit(1);
- }
-}
-//**********************************************************************************************************************
-vector<string> DeconvoluteCommand::getRequiredParameters(){
- try {
- string Array[] = {"fasta"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
+ string helpString = "";
+ helpString += "The unique.seqs command reads a fastafile and creates a namesfile.\n";
+ helpString += "It creates a file where the first column is the groupname and the second column is a list of sequence names who have the same sequence. \n";
+ helpString += "If the sequence is unique the second column will just contain its name. \n";
+ helpString += "The unique.seqs command parameters are fasta and name. fasta is required, unless there is a valid current fasta file.\n";
+ helpString += "The unique.seqs command should be in the following format: \n";
+ helpString += "unique.seqs(fasta=yourFastaFile) \n";
+ return helpString;
}
catch(exception& e) {
- m->errorOut(e, "DeconvoluteCommand", "getRequiredParameters");
+ m->errorOut(e, "DeconvoluteCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
-vector<string> DeconvoluteCommand::getRequiredFiles(){
+DeconvoluteCommand::DeconvoluteCommand(){
try {
- vector<string> myArray;
- return myArray;
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["name"] = tempOutNames;
}
catch(exception& e) {
- m->errorOut(e, "DeconvoluteCommand", "getRequiredFiles");
+ m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
exit(1);
}
}
/**************************************************************************************/
DeconvoluteCommand::DeconvoluteCommand(string option) {
try {
- abort = false;
+ abort = false; calledHelp = false;
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string Array[] = {"fasta", "name","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
//check for required parameters
inFastaName = validParameter.validFile(parameters, "fasta", true);
if (inFastaName == "not open") { abort = true; }
- else if (inFastaName == "not found") { inFastaName = ""; m->mothurOut("fasta is a required parameter for the unique.seqs command."); m->mothurOutEndLine(); abort = true; }
+ else if (inFastaName == "not found") {
+ inFastaName = m->getFastaFile();
+ if (inFastaName != "") { m->mothurOut("Using " + inFastaName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }else { m->setFastaFile(inFastaName); }
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
oldNameMapFName = validParameter.validFile(parameters, "name", true);
if (oldNameMapFName == "not open") { abort = true; }
else if (oldNameMapFName == "not found"){ oldNameMapFName = ""; }
+ else { m->setNameFile(oldNameMapFName); }
}
}
exit(1);
}
}
-//**********************************************************************************************************************
-
-void DeconvoluteCommand::help(){
- try {
- m->mothurOut("The unique.seqs command reads a fastafile and creates a namesfile.\n");
- m->mothurOut("It creates a file where the first column is the groupname and the second column is a list of sequence names who have the same sequence. \n");
- m->mothurOut("If the sequence is unique the second column will just contain its name. \n");
- m->mothurOut("The unique.seqs command parameter is fasta and it is required.\n");
- m->mothurOut("The unique.seqs command should be in the following format: \n");
- m->mothurOut("unique.seqs(fasta=yourFastaFile) \n");
- }
- catch(exception& e) {
- m->errorOut(e, "DeconvoluteCommand", "help");
- exit(1);
- }
-}
-
/**************************************************************************************/
int DeconvoluteCommand::execute() {
try {
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
//prepare filenames and open files
string outNameFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "names";
string outFastaFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "unique" + m->getExtension(inFastaName);
- FastaMap fastamap;
-
- if(oldNameMapFName == "") { fastamap.readFastaFile(inFastaName); }
- else { fastamap.readFastaFile(inFastaName, oldNameMapFName); }
+ map<string, string> nameMap;
+ map<string, string>::iterator itNames;
+ if (oldNameMapFName != "") { m->readNames(oldNameMapFName, nameMap); }
if (m->control_pressed) { return 0; }
- fastamap.printCondensedFasta(outFastaFile);
- fastamap.printNamesFile(outNameFile);
+ ifstream in;
+ m->openInputFile(inFastaName, in);
+
+ ofstream outFasta;
+ m->openOutputFile(outFastaFile, outFasta);
+
+ map<string, string> sequenceStrings; //sequenceString -> list of names. "atgc...." -> seq1,seq2,seq3.
+ map<string, string>::iterator itStrings;
+ set<string> nameInFastaFile; //for sanity checking
+ set<string>::iterator itname;
+ vector<string> nameFileOrder;
+ int count = 0;
+ while (!in.eof()) {
+
+ if (m->control_pressed) { in.close(); outFasta.close(); m->mothurRemove(outFastaFile); return 0; }
+
+ Sequence seq(in);
+
+ if (seq.getName() != "") {
+
+ //sanity checks
+ itname = nameInFastaFile.find(seq.getName());
+ if (itname == nameInFastaFile.end()) { nameInFastaFile.insert(seq.getName()); }
+ else { m->mothurOut("[ERROR]: You already have a sequence named " + seq.getName() + " in your fasta file, sequence names must be unique, please correct."); m->mothurOutEndLine(); }
+
+ itStrings = sequenceStrings.find(seq.getAligned());
+
+ if (itStrings == sequenceStrings.end()) { //this is a new unique sequence
+ //output to unique fasta file
+ seq.printSequence(outFasta);
+
+ if (oldNameMapFName != "") {
+ itNames = nameMap.find(seq.getName());
+
+ if (itNames == nameMap.end()) { //namefile and fastafile do not match
+ m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
+ }else {
+ sequenceStrings[seq.getAligned()] = itNames->second;
+ nameFileOrder.push_back(seq.getAligned());
+ }
+ }else { sequenceStrings[seq.getAligned()] = seq.getName(); nameFileOrder.push_back(seq.getAligned()); }
+ }else { //this is a dup
+ if (oldNameMapFName != "") {
+ itNames = nameMap.find(seq.getName());
+
+ if (itNames == nameMap.end()) { //namefile and fastafile do not match
+ m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
+ }else {
+ sequenceStrings[seq.getAligned()] += "," + itNames->second;
+ }
+ }else { sequenceStrings[seq.getAligned()] += "," + seq.getName(); }
+ }
+
+ count++;
+ }
+
+ m->gobble(in);
+
+ if(count % 1000 == 0) { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine(); }
+ }
- if (m->control_pressed) { outputTypes.clear(); remove(outFastaFile.c_str()); remove(outNameFile.c_str()); return 0; }
+ if(count % 1000 != 0) { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine(); }
+
+ in.close();
+ outFasta.close();
+
+ if (m->control_pressed) { m->mothurRemove(outFastaFile); return 0; }
+
+ //print new names file
+ ofstream outNames;
+ m->openOutputFile(outNameFile, outNames);
+
+ for (int i = 0; i < nameFileOrder.size(); i++) {
+ //for (itStrings = sequenceStrings.begin(); itStrings != sequenceStrings.end(); itStrings++) {
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); outNames.close(); m->mothurRemove(outNameFile); return 0; }
+
+ itStrings = sequenceStrings.find(nameFileOrder[i]);
+
+ if (itStrings != sequenceStrings.end()) {
+ //get rep name
+ int pos = (itStrings->second).find_first_of(',');
+
+ if (pos == string::npos) { // only reps itself
+ outNames << itStrings->second << '\t' << itStrings->second << endl;
+ }else {
+ outNames << (itStrings->second).substr(0, pos) << '\t' << itStrings->second << endl;
+ }
+ }else{ m->mothurOut("[ERROR]: mismatch in namefile print."); m->mothurOutEndLine(); m->control_pressed = true; }
+ }
+ outNames.close();
+
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); m->mothurRemove(outNameFile); return 0; }
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
outputNames.push_back(outFastaFile); outputNames.push_back(outNameFile); outputTypes["fasta"].push_back(outFastaFile); outputTypes["name"].push_back(outNameFile);
m->mothurOutEndLine();
-
+ //set fasta file as new current fastafile
+ string current = "";
+ itTypes = outputTypes.find("fasta");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
+ }
+
+ itTypes = outputTypes.find("name");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
+ }
return 0;
}
projects / mothur.git / blobdiff