#include "listvector.hpp"
//**********************************************************************************************************************
-
+vector<string> RemoveSeqsCommand::getValidParameters(){
+ try {
+ string Array[] = {"fasta","name", "group", "alignreport", "accnos", "qfile","list","taxonomy","outputdir","inputdir", "dups" };
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "getValidParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+RemoveSeqsCommand::RemoveSeqsCommand(){
+ try {
+ abort = true; calledHelp = true;
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["taxonomy"] = tempOutNames;
+ outputTypes["name"] = tempOutNames;
+ outputTypes["group"] = tempOutNames;
+ outputTypes["alignreport"] = tempOutNames;
+ outputTypes["list"] = tempOutNames;
+ outputTypes["qfile"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> RemoveSeqsCommand::getRequiredParameters(){
+ try {
+ string Array[] = {"accnos"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "getRequiredParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> RemoveSeqsCommand::getRequiredFiles(){
+ try {
+ vector<string> myArray;
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "getRequiredFiles");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
RemoveSeqsCommand::RemoveSeqsCommand(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 {
//valid paramters for this command
- string Array[] = {"fasta","name", "group", "alignreport", "accnos", "list","outputdir","inputdir", "dups" };
+ string Array[] = {"fasta","name", "group", "alignreport", "accnos", "qfile", "list","taxonomy","outputdir","inputdir", "dups" };
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
OptionParser parser(option);
if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
}
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["taxonomy"] = tempOutNames;
+ outputTypes["name"] = tempOutNames;
+ outputTypes["group"] = tempOutNames;
+ outputTypes["alignreport"] = tempOutNames;
+ outputTypes["list"] = tempOutNames;
+ outputTypes["qfile"] = tempOutNames;
+
//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"){ outputDir = ""; }
it = parameters.find("alignreport");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["alignreport"] = inputDir + it->second; }
}
it = parameters.find("fasta");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["fasta"] = inputDir + it->second; }
}
it = parameters.find("accnos");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["accnos"] = inputDir + it->second; }
}
it = parameters.find("list");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["list"] = inputDir + it->second; }
}
it = parameters.find("name");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["name"] = inputDir + it->second; }
}
it = parameters.find("group");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["group"] = inputDir + it->second; }
}
+
+ it = parameters.find("taxonomy");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["taxonomy"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("qfile");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["qfile"] = inputDir + it->second; }
+ }
}
if (listfile == "not open") { abort = true; }
else if (listfile == "not found") { listfile = ""; }
- string usedDups = "true";
- string temp = validParameter.validFile(parameters, "dups", false); if (temp == "not found") { temp = "false"; usedDups = ""; }
- dups = isTrue(temp);
+ taxfile = validParameter.validFile(parameters, "taxonomy", true);
+ if (taxfile == "not open") { abort = true; }
+ else if (taxfile == "not found") { taxfile = ""; }
- if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "")) { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport or list."); m->mothurOutEndLine(); abort = true; }
+ qualfile = validParameter.validFile(parameters, "qfile", true);
+ if (qualfile == "not open") { abort = true; }
+ else if (qualfile == "not found") { qualfile = ""; }
+
- //int okay = 2;
- //if (outputDir != "") { okay++; }
- //if (usedDups != "") { okay++; }
+ string usedDups = "true";
+ string temp = validParameter.validFile(parameters, "dups", false);
+ if (temp == "not found") {
+ if (namefile != "") { temp = "true"; }
+ else { temp = "false"; usedDups = ""; }
+ }
+ dups = m->isTrue(temp);
- if ((usedDups != "") && (namefile == "")) { m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine(); abort = true; }
+ if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == "")) { m->mothurOut("You must provide at least one of the following: fasta, name, group, taxonomy, quality, alignreport or list."); m->mothurOutEndLine(); abort = true; }
- //if (parameters.size() > okay) { m->mothurOut("You may only enter one of the following: fasta, name, group, alignreport, or list."); m->mothurOutEndLine(); abort = true; }
+ if ((usedDups != "") && (namefile == "")) { m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine(); abort = true; }
}
}
void RemoveSeqsCommand::help(){
try {
- m->mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
+ m->mothurOut("The remove.seqs command reads an .accnos file and at least one of the following file types: fasta, name, group, list, taxonomy, quality or alignreport file.\n");
m->mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
- m->mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list, alignreport and dups. You must provide accnos and one of the file parameters.\n");
- m->mothurOut("The dups parameter allows you to remove the entire line from a name file if you remove any name from the line. default=false. If dups=true, then remove.seqs outputs a new .accnos file containing all the sequences removed. \n");
+ m->mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups. You must provide accnos and at least one of the file parameters.\n");
+ m->mothurOut("The dups parameter allows you to remove the entire line from a name file if you remove any name from the line. default=true. \n");
m->mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
int RemoveSeqsCommand::execute(){
try {
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
//get names you want to keep
readAccnos();
if (m->control_pressed) { return 0; }
//read through the correct file and output lines you want to keep
- if (fastafile != "") { readFasta(); }
if (namefile != "") { readName(); }
+ if (fastafile != "") { readFasta(); }
if (groupfile != "") { readGroup(); }
if (alignfile != "") { readAlign(); }
if (listfile != "") { readList(); }
+ if (taxfile != "") { readTax(); }
+ if (qualfile != "") { readQual(); }
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
+ m->mothurOut("Removed " + toString(names.size()) + " sequences."); m->mothurOutEndLine();
+
if (outputNames.size() != 0) {
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
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); }
+ }
+
+ itTypes = outputTypes.find("group");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
+ }
+
+ itTypes = outputTypes.find("list");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
+ }
+
+ itTypes = outputTypes.find("taxonomy");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
+ }
+
+ itTypes = outputTypes.find("qfile");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
+ }
}
return 0;
//**********************************************************************************************************************
int RemoveSeqsCommand::readFasta(){
try {
- if (outputDir == "") { outputDir += hasPath(fastafile); }
- string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pick" + getExtension(fastafile);
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(fastafile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
+
ofstream out;
- openOutputFile(outputFileName, out);
+ m->openOutputFile(outputFileName, out);
ifstream in;
- openInputFile(fastafile, in);
+ m->openInputFile(fastafile, in);
string name;
bool wroteSomething = false;
wroteSomething = true;
currSeq.printSequence(out);
- }//else { names.erase(name); }
+ }
}
- gobble(in);
+ m->gobble(in);
}
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["fasta"].push_back(outputFileName); outputNames.push_back(outputFileName);
return 0;
}
}
//**********************************************************************************************************************
+int RemoveSeqsCommand::readQual(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(qualfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" + m->getExtension(qualfile);
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+
+ ifstream in;
+ m->openInputFile(qualfile, in);
+ string name;
+
+ bool wroteSomething = false;
+
+
+ while(!in.eof()){
+ string saveName = "";
+ string name = "";
+ string scores = "";
+
+ in >> name;
+
+ if (name.length() != 0) {
+ saveName = name.substr(1);
+ while (!in.eof()) {
+ char c = in.get();
+ if (c == 10 || c == 13){ break; }
+ else { name += c; }
+ }
+ m->gobble(in);
+ }
+
+ while(in){
+ char letter= in.get();
+ if(letter == '>'){ in.putback(letter); break; }
+ else{ scores += letter; }
+ }
+
+ m->gobble(in);
+
+ if (names.count(saveName) == 0) {
+ wroteSomething = true;
+
+ out << name << endl << scores;
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName); outputTypes["qfile"].push_back(outputFileName);
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "readQual");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
int RemoveSeqsCommand::readList(){
try {
- if (outputDir == "") { outputDir += hasPath(listfile); }
- string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + "pick" + getExtension(listfile);
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
+
ofstream out;
- openOutputFile(outputFileName, out);
+ m->openOutputFile(outputFileName, out);
ifstream in;
- openInputFile(listfile, in);
+ m->openInputFile(listfile, in);
bool wroteSomething = false;
newList.print(out);
}
- gobble(in);
+ m->gobble(in);
}
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
-
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
return 0;
}
//**********************************************************************************************************************
int RemoveSeqsCommand::readName(){
try {
- if (outputDir == "") { outputDir += hasPath(namefile); }
- string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" + getExtension(namefile);
- string outputFileName2 = outputDir + getRootName(getSimpleName(namefile)) + "dups.accnos";
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(namefile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
- ofstream out2;
- if (dups) { openOutputFile(outputFileName2, out2); }
- bool wroteDups = false;
-
ofstream out;
- openOutputFile(outputFileName, out);
+ m->openOutputFile(outputFileName, out);
ifstream in;
- openInputFile(namefile, in);
+ m->openInputFile(namefile, in);
string name, firstCol, secondCol;
bool wroteSomething = false;
+
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); if (dups) { out2.close(); remove(outputFileName2.c_str()); } return 0; }
-
- in >> firstCol;
+ if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+
+ in >> firstCol; m->gobble(in);
in >> secondCol;
-
+
vector<string> parsedNames;
- //parse second column saving each name
- while (secondCol.find_first_of(',') != -1) {
- name = secondCol.substr(0,secondCol.find_first_of(','));
- secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
- parsedNames.push_back(name);
-
- }
+ m->splitAtComma(secondCol, parsedNames);
- //get name after last ,
- parsedNames.push_back(secondCol);
-
vector<string> validSecond; validSecond.clear();
for (int i = 0; i < parsedNames.size(); i++) {
if (names.count(parsedNames[i]) == 0) {
}
}
- if ((dups) && (validSecond.size() != parsedNames.size())) {
- wroteDups = true;
- for (int i = 0; i < parsedNames.size(); i++) { out2 << parsedNames[i] << endl; }
+ if ((dups) && (validSecond.size() != parsedNames.size())) { //if dups is true and we want to get rid of anyone, get rid of everyone
+ for (int i = 0; i < parsedNames.size(); i++) { names.insert(parsedNames[i]); }
}else {
- //if the name in the first column is in the set then print it and any other names in second column also in set
+ //if the name in the first column is in the set then print it and any other names in second column also in set
if (names.count(firstCol) == 0) {
wroteSomething = true;
}
}
}
- gobble(in);
+ m->gobble(in);
}
in.close();
out.close();
- if (dups) { out2.close(); }
- if (wroteDups == false) {
- remove(outputFileName2.c_str());
- }else { outputNames.push_back(outputFileName2); }
-
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
return 0;
}
//**********************************************************************************************************************
int RemoveSeqsCommand::readGroup(){
try {
- if (outputDir == "") { outputDir += hasPath(groupfile); }
- string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(groupfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
+
ofstream out;
- openOutputFile(outputFileName, out);
+ m->openOutputFile(outputFileName, out);
ifstream in;
- openInputFile(groupfile, in);
+ m->openInputFile(groupfile, in);
string name, group;
bool wroteSomething = false;
if (names.count(name) == 0) {
wroteSomething = true;
out << name << '\t' << group << endl;
- }//else { names.erase(name); }
+ }
- gobble(in);
+ m->gobble(in);
}
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
return 0;
}
exit(1);
}
}
+//**********************************************************************************************************************
+int RemoveSeqsCommand::readTax(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(taxfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+ ifstream in;
+ m->openInputFile(taxfile, in);
+ string name, tax;
+
+ bool wroteSomething = false;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+
+ in >> name; //read from first column
+ in >> tax; //read from second column
+
+ //if this name is in the accnos file
+ if (names.count(name) == 0) {
+ wroteSomething = true;
+ out << name << '\t' << tax << endl;
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "readTax");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
//alignreport file has a column header line then all other lines contain 16 columns. we just want the first column since that contains the name
int RemoveSeqsCommand::readAlign(){
try {
- if (outputDir == "") { outputDir += hasPath(alignfile); }
- string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(alignfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
+
ofstream out;
- openOutputFile(outputFileName, out);
+ m->openOutputFile(outputFileName, out);
ifstream in;
- openInputFile(alignfile, in);
+ m->openInputFile(alignfile, in);
string name, junk;
bool wroteSomething = false;
out << endl;
}else {//still read just don't do anything with it
- //names.erase(name);
//read rest
for (int i = 0; i < 15; i++) {
}
}
- gobble(in);
+ m->gobble(in);
}
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
return 0;
try {
ifstream in;
- openInputFile(accnosfile, in);
+ m->openInputFile(accnosfile, in);
string name;
while(!in.eof()){
names.insert(name);
- gobble(in);
+ m->gobble(in);
}
in.close();
projects / mothur.git / blobdiff