vector<string> RemoveSeqsCommand::setParameters(){
try {
CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
- CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
- CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pgroup);
+ CommandParameter pname("name", "InputTypes", "", "", "NameCount", "FNGLT", "none",false,false); parameters.push_back(pname);
+ CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "FNGLT", "none",false,false); parameters.push_back(pcount);
+ CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "FNGLT", "none",false,false); parameters.push_back(pgroup);
CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
string RemoveSeqsCommand::getHelpString(){
try {
string helpString = "";
- helpString += "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";
+ helpString += "The remove.seqs command reads an .accnos file and at least one of the following file types: fasta, name, group, count, list, taxonomy, quality or alignreport file.\n";
helpString += "It outputs a file containing the sequences NOT in the .accnos file.\n";
- helpString += "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";
+ helpString += "The remove.seqs command parameters are accnos, fasta, name, group, count, list, taxonomy, qfile, alignreport and dups. You must provide accnos and at least one of the file parameters.\n";
helpString += "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";
helpString += "The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
helpString += "Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
exit(1);
}
}
-
+//**********************************************************************************************************************
+string RemoveSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
+ try {
+ string outputFileName = "";
+ map<string, vector<string> >::iterator it;
+
+ //is this a type this command creates
+ it = outputTypes.find(type);
+ if (it == outputTypes.end()) { m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+ else {
+ if (type == "fasta") { outputFileName = "pick" + m->getExtension(inputName); }
+ else if (type == "taxonomy") { outputFileName = "pick" + m->getExtension(inputName); }
+ else if (type == "name") { outputFileName = "pick" + m->getExtension(inputName); }
+ else if (type == "group") { outputFileName = "pick" + m->getExtension(inputName); }
+ else if (type == "list") { outputFileName = "pick" + m->getExtension(inputName); }
+ else if (type == "qfile") { outputFileName = "pick" + m->getExtension(inputName); }
+ else if (type == "alignreport") { outputFileName = "pick.align.report"; }
+ else if (type == "count") { outputFileName = "pick.count.table"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return outputFileName;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "getOutputFileNameTag");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
RemoveSeqsCommand::RemoveSeqsCommand(){
outputTypes["alignreport"] = tempOutNames;
outputTypes["list"] = tempOutNames;
outputTypes["qfile"] = tempOutNames;
+ outputTypes["count"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
outputTypes["alignreport"] = tempOutNames;
outputTypes["list"] = tempOutNames;
outputTypes["qfile"] = tempOutNames;
+ outputTypes["count"] = 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 = ""; }
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["qfile"] = inputDir + it->second; }
}
+
+ it = parameters.find("count");
+ //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["count"] = inputDir + it->second; }
+ }
}
m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
abort = true;
}
- }
+ }else { m->setAccnosFile(accnosfile); }
fastafile = validParameter.validFile(parameters, "fasta", true);
- if (fastafile == "not open") { abort = true; }
+ if (fastafile == "not open") { fastafile = ""; abort = true; }
else if (fastafile == "not found") { fastafile = ""; }
-
+ else { m->setFastaFile(fastafile); }
+
namefile = validParameter.validFile(parameters, "name", true);
- if (namefile == "not open") { abort = true; }
+ if (namefile == "not open") { namefile = ""; abort = true; }
else if (namefile == "not found") { namefile = ""; }
-
+ else { m->setNameFile(namefile); }
+
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { abort = true; }
- else if (groupfile == "not found") { groupfile = ""; }
+ else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
alignfile = validParameter.validFile(parameters, "alignreport", true);
if (alignfile == "not open") { abort = true; }
listfile = validParameter.validFile(parameters, "list", true);
if (listfile == "not open") { abort = true; }
else if (listfile == "not found") { listfile = ""; }
+ else { m->setListFile(listfile); }
taxfile = validParameter.validFile(parameters, "taxonomy", true);
if (taxfile == "not open") { abort = true; }
else if (taxfile == "not found") { taxfile = ""; }
+ else { m->setTaxonomyFile(taxfile); }
qualfile = validParameter.validFile(parameters, "qfile", true);
if (qualfile == "not open") { abort = true; }
else if (qualfile == "not found") { qualfile = ""; }
-
+ else { m->setQualFile(qualfile); }
string usedDups = "true";
string temp = validParameter.validFile(parameters, "dups", false);
else { temp = "false"; usedDups = ""; }
}
dups = m->isTrue(temp);
-
- 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; }
-
+
+ countfile = validParameter.validFile(parameters, "count", true);
+ if (countfile == "not open") { countfile = ""; abort = true; }
+ else if (countfile == "not found") { countfile = ""; }
+ else { m->setCountTableFile(countfile); }
+
+ if ((namefile != "") && (countfile != "")) {
+ m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
+ }
+
+ if ((groupfile != "") && (countfile != "")) {
+ m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
+ }
+
+ if ((countfile == "") && (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 (countfile == "") {
+ if ((fastafile != "") && (namefile == "")) {
+ vector<string> files; files.push_back(fastafile);
+ parser.getNameFile(files);
+ }
+ }
}
}
if (abort == true) { if (calledHelp) { return 0; } return 2; }
//get names you want to keep
- readAccnos();
+ names = m->readAccnos(accnosfile);
if (m->control_pressed) { return 0; }
+
+ if (countfile != "") {
+ if ((fastafile != "") || (listfile != "") || (taxfile != "")) {
+ m->mothurOut("\n[NOTE]: The count file should contain only unique names, so mothur assumes your fasta, list and taxonomy files also contain only uniques.\n\n");
+ }
+ }
//read through the correct file and output lines you want to keep
if (namefile != "") { readName(); }
if (listfile != "") { readList(); }
if (taxfile != "") { readTax(); }
if (qualfile != "") { readQual(); }
+ if (countfile != "") { readCount(); }
- 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 (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
+
if (outputNames.size() != 0) {
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
itTypes = outputTypes.find("qfile");
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
- }
+ }
+
+ itTypes = outputTypes.find("count");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
+ }
}
return 0;
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(fastafile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta", fastafile);
ofstream out;
m->openOutputFile(outputFileName, out);
string name;
bool wroteSomething = false;
+ int removedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
Sequence currSeq(in);
name = currSeq.getName();
wroteSomething = true;
currSeq.printSequence(out);
- }
+ }else { removedCount++; }
}
m->gobble(in);
}
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);
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
+
return 0;
}
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(qualfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" + m->getExtension(qualfile);
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + getOutputFileNameTag("qfile", qualfile);
ofstream out;
m->openOutputFile(outputFileName, out);
string name;
bool wroteSomething = false;
+ int removedCount = 0;
while(!in.eof()){
wroteSomething = true;
out << name << endl << scores;
- }
+ }else { removedCount++; }
m->gobble(in);
}
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);
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your quality file."); m->mothurOutEndLine();
+
return 0;
}
}
}
//**********************************************************************************************************************
+int RemoveSeqsCommand::readCount(){
+ try {
+
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(countfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(countfile)) + getOutputFileNameTag("count", countfile);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(countfile, in);
+
+ bool wroteSomething = false;
+ int removedCount = 0;
+
+ string headers = m->getline(in); m->gobble(in);
+ out << headers << endl;
+
+ string name, rest; int thisTotal;
+ while (!in.eof()) {
+
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ in >> name; m->gobble(in);
+ in >> thisTotal; m->gobble(in);
+ rest = m->getline(in); m->gobble(in);
+ if (m->debug) { m->mothurOut("[DEBUG]: " + name + '\t' + rest + "\n"); }
+
+ if (names.count(name) == 0) {
+ out << name << '\t' << thisTotal << '\t' << rest << endl;
+ wroteSomething = true;
+ }else { removedCount += thisTotal; }
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your count file."); m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "readCount");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
int RemoveSeqsCommand::readList(){
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + getOutputFileNameTag("list", listfile);
ofstream out;
m->openOutputFile(outputFileName, out);
m->openInputFile(listfile, in);
bool wroteSomething = false;
+ int removedCount = 0;
while(!in.eof()){
+
+ removedCount = 0;
+
//read in list vector
ListVector list(in);
//for each bin
for (int i = 0; i < list.getNumBins(); i++) {
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
//parse out names that are in accnos file
string binnames = list.get(i);
//if that name is in the .accnos file, add it
if (names.count(name) == 0) { newNames += name + ","; }
+ else { removedCount++; }
}
//get last name
if (names.count(binnames) == 0) { newNames += binnames + ","; }
+ else { removedCount++; }
//if there are names in this bin add to new list
if (newNames != "") {
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);
-
+
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
+
return 0;
}
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(namefile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + getOutputFileNameTag("name", namefile);
ofstream out;
m->openOutputFile(outputFileName, out);
string name, firstCol, secondCol;
bool wroteSomething = false;
-
+ int removedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> firstCol; m->gobble(in);
in >> secondCol;
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]); }
+ removedCount += parsedNames.size();
}else {
+ removedCount += parsedNames.size()-validSecond.size();
//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) {
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);
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
+
return 0;
}
catch(exception& e) {
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(groupfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
ofstream out;
m->openOutputFile(outputFileName, out);
string name, group;
bool wroteSomething = false;
+ int removedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> name; //read from first column
in >> group; //read from second column
if (names.count(name) == 0) {
wroteSomething = true;
out << name << '\t' << group << endl;
- }
+ }else { removedCount++; }
m->gobble(in);
}
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);
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
+
+
return 0;
}
catch(exception& e) {
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(taxfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + getOutputFileNameTag("taxonomy", taxfile);
ofstream out;
m->openOutputFile(outputFileName, out);
string name, tax;
bool wroteSomething = false;
+ int removedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> name; //read from first column
in >> tax; //read from second column
if (names.count(name) == 0) {
wroteSomething = true;
out << name << '\t' << tax << endl;
- }
+ }else { removedCount++; }
m->gobble(in);
}
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);
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
+
return 0;
}
catch(exception& e) {
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(alignfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + getOutputFileNameTag("alignreport");
ofstream out;
m->openOutputFile(outputFileName, out);
string name, junk;
bool wroteSomething = false;
+ int removedCount = 0;
//read column headers
for (int i = 0; i < 16; i++) {
out << endl;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> name; //read from first column
out << endl;
}else {//still read just don't do anything with it
+ removedCount++;
//read rest
for (int i = 0; i < 15; i++) {
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);
+ m->mothurOut("Removed " + toString(removedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
+
+
return 0;
}
}
}
//**********************************************************************************************************************
-void RemoveSeqsCommand::readAccnos(){
- try {
-
- ifstream in;
- m->openInputFile(accnosfile, in);
- string name;
-
- while(!in.eof()){
- in >> name;
-
- names.insert(name);
-
- m->gobble(in);
- }
- in.close();
-
- }
- catch(exception& e) {
- m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
projects / mothur.git / blobdiff