#include "getlineagecommand.h"
#include "sequence.hpp"
#include "listvector.hpp"
+#include "counttable.h"
//**********************************************************************************************************************
vector<string> GetLineageCommand::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 plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
- CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,true); parameters.push_back(ptaxonomy);
- CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
- CommandParameter ptaxon("taxon", "String", "", "", "", "", "",false,true); parameters.push_back(ptaxon);
- CommandParameter pdups("dups", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pdups);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none","fasta",false,false, true); parameters.push_back(pfasta);
+ CommandParameter pname("name", "InputTypes", "", "", "NameCount", "FNGLT", "none","name",false,false, true); parameters.push_back(pname);
+ CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "FNGLT", "none","count",false,false, true); parameters.push_back(pcount);
+ CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "FNGLT", "none","group",false,false, true); parameters.push_back(pgroup);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none","list",false,false, true); parameters.push_back(plist);
+ CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none","taxonomy",false,true, true); parameters.push_back(ptaxonomy);
+ CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none","alignreport",false,false); parameters.push_back(palignreport);
+ CommandParameter ptaxon("taxon", "String", "", "", "", "", "","",false,true, true); parameters.push_back(ptaxon);
+ CommandParameter pdups("dups", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pdups);
+ 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); }
string GetLineageCommand::getHelpString(){
try {
string helpString = "";
- helpString += "The get.lineage command reads a taxonomy file and any of the following file types: fasta, name, group, list or alignreport file.\n";
+ helpString += "The get.lineage command reads a taxonomy file and any of the following file types: fasta, name, group, count, list or alignreport file.\n";
helpString += "It outputs a file containing only the sequences from the taxonomy file that are from the taxon requested.\n";
- helpString += "The get.lineage command parameters are taxon, fasta, name, group, list, taxonomy, alignreport and dups. You must provide taxonomy unless you have a valid current taxonomy file.\n";
+ helpString += "The get.lineage command parameters are taxon, fasta, name, group, count, list, taxonomy, alignreport and dups. You must provide taxonomy unless you have a valid current taxonomy file.\n";
helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=false. \n";
helpString += "The taxon parameter allows you to select the taxons you would like to get and is required.\n";
helpString += "You may enter your taxons with confidence scores, doing so will get only those sequences that belong to the taxonomy and whose cofidence scores is above the scores you give.\n";
exit(1);
}
}
-
+//**********************************************************************************************************************
+string GetLineageCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "fasta") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "taxonomy") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "name") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "group") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "count") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "list") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "alignreport") { pattern = "[filename],pick.align.report"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetLineageCommand", "getOutputPattern");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
GetLineageCommand::GetLineageCommand(){
try {
outputTypes["group"] = tempOutNames;
outputTypes["alignreport"] = tempOutNames;
outputTypes["list"] = tempOutNames;
+ outputTypes["count"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
outputTypes["group"] = tempOutNames;
outputTypes["alignreport"] = tempOutNames;
outputTypes["list"] = 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["taxonomy"] = 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; }
+ }
}
//check for required parameters
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); }
else { m->setListFile(listfile); }
taxfile = validParameter.validFile(parameters, "taxonomy", true);
- if (taxfile == "not open") { abort = true; }
+ if (taxfile == "not open") { taxfile = ""; abort = true; }
else if (taxfile == "not found") {
taxfile = m->getTaxonomyFile();
if (taxfile != "") { m->mothurOut("Using " + taxfile + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
else { temp = "false"; usedDups = ""; }
}
dups = m->isTrue(temp);
+
+ 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;
+ }
taxons = validParameter.validFile(parameters, "taxon", false);
if (taxons == "not found") { taxons = ""; m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine(); abort = true; }
}
m->splitAtChar(taxons, listOfTaxons, '-');
- if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "")) { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport, taxonomy or listfile."); m->mothurOutEndLine(); abort = true; }
+ if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (countfile == "")) { m->mothurOut("You must provide one of the following: fasta, name, group, count, alignreport, taxonomy or listfile."); m->mothurOutEndLine(); abort = true; }
+
+ if (countfile == "") {
+ if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
+ vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
+ parser.getNameFile(files);
+ }
+ }
}
}
if (abort == true) { if (calledHelp) { return 0; } return 2; }
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 (taxfile != "") { readTax(); } //fills the set of names to get
if (namefile != "") { readName(); }
if (fastafile != "") { readFasta(); }
+ if (countfile != "") { readCount(); }
if (groupfile != "") { readGroup(); }
if (alignfile != "") { readAlign(); }
if (listfile != "") { readList(); }
- if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
+ if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
if (outputNames.size() != 0) {
m->mothurOutEndLine();
itTypes = outputTypes.find("taxonomy");
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(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);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(fastafile));
+ variables["[extension]"] = m->getExtension(fastafile);
+ string outputFileName = getOutputFileName("fasta", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
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();
in.close();
out.close();
- if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains does not contain any sequences from " + taxons + "."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
return 0;
}
}
//**********************************************************************************************************************
+int GetLineageCommand::readCount(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(countfile); }
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
+ variables["[extension]"] = m->getExtension(countfile);
+ string outputFileName = getOutputFileName("count", variables);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(countfile, in);
+
+ bool wroteSomething = false;
+
+ 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;
+ }
+ }
+ in.close();
+ out.close();
+
+ //check for groups that have been eliminated
+ CountTable ct;
+ if (ct.testGroups(outputFileName)) {
+ ct.readTable(outputFileName);
+ ct.printTable(outputFileName);
+ }
+
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains does not contain any sequences from " + taxons + "."); m->mothurOutEndLine(); }
+ outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetLineageCommand", "readCount");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
int GetLineageCommand::readList(){
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
+ variables["[extension]"] = m->getExtension(listfile);
+ string outputFileName = getOutputFileName("list", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
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; }
//read in list vector
ListVector list(in);
in.close();
out.close();
- if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains does not contain any sequences from " + taxons + "."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
return 0;
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(namefile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(namefile));
+ variables["[extension]"] = m->getExtension(namefile);
+ string outputFileName = getOutputFileName("name", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
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;
in >> secondCol;
in.close();
out.close();
- if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains does not contain any sequences from " + taxons + "."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
return 0;
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(groupfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(groupfile));
+ variables["[extension]"] = m->getExtension(groupfile);
+ string outputFileName = getOutputFileName("group", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
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.close();
out.close();
- if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains does not contain any sequences from " + taxons + "."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["group"].push_back(outputFileName);
return 0;
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(taxfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(taxfile));
+ variables["[extension]"] = m->getExtension(taxfile);
+ string outputFileName = getOutputFileName("taxonomy", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
if (hasConPos != string::npos) {
taxonsHasConfidence[i] = true;
searchTaxons[i] = getTaxons(listOfTaxons[i]);
- noConfidenceTaxons[i] = removeConfidences(listOfTaxons[i]);
+ noConfidenceTaxons[i] = listOfTaxons[i];
+ m->removeConfidences(noConfidenceTaxons[i]);
}
}
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
+ string noQuotesTax = m->removeQuotes(tax);
+
for (int j = 0; j < listOfTaxons.size(); j++) {
- string newtax = tax;
+ string newtax = noQuotesTax;
//if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
if (!taxonsHasConfidence[j]) {
- int hasConfidences = tax.find_first_of('(');
+ int hasConfidences = noQuotesTax.find_first_of('(');
if (hasConfidences != string::npos) {
- newtax = removeConfidences(tax);
+ newtax = noQuotesTax;
+ m->removeConfidences(newtax);
}
int pos = newtax.find(noConfidenceTaxons[j]);
break;
}
}else{//if listOfTaxons[i] has them and you don't them remove taxons
- int hasConfidences = tax.find_first_of('(');
+ int hasConfidences = noQuotesTax.find_first_of('(');
if (hasConfidences == string::npos) {
int pos = newtax.find(noConfidenceTaxons[j]);
}else { //both have confidences so we want to make sure the users confidences are greater then or equal to the taxons
//first remove confidences from both and see if the taxonomy exists
- string noNewTax = tax;
- int hasConfidences = tax.find_first_of('(');
+ string noNewTax = noQuotesTax;
+ int hasConfidences = noQuotesTax.find_first_of('(');
if (hasConfidences != string::npos) {
- noNewTax = removeConfidences(tax);
+ noNewTax = noQuotesTax;
+ m->removeConfidences(noNewTax);
}
int pos = noNewTax.find(noConfidenceTaxons[j]);
/**************************************************************************************************/
vector< map<string, float> > GetLineageCommand::getTaxons(string tax) {
try {
-
+
vector< map<string, float> > t;
string taxon = "";
int taxLength = tax.length();
+
for(int i=0;i<taxLength;i++){
if(tax[i] == ';'){
-
- int openParen = taxon.find_first_of('(');
+
+ int openParen = taxon.find_last_of('(');
int closeParen = taxon.find_last_of(')');
string newtaxon, confidence;
if ((openParen != string::npos) && (closeParen != string::npos)) {
- newtaxon = taxon.substr(0, openParen); //rip off confidence
- confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
+ string confidenceScore = taxon.substr(openParen+1, (closeParen-(openParen+1)));
+ if (m->isNumeric1(confidenceScore)) { //its a confidence
+ newtaxon = taxon.substr(0, openParen); //rip off confidence
+ confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
+ }else { //its part of the taxon
+ newtaxon = taxon;
+ confidence = "0";
+ }
}else{
newtaxon = taxon;
confidence = "0";
map<string, float> temp;
temp[newtaxon] = con;
+
t.push_back(temp);
-
taxon = "";
}
else{
taxon += tax[i];
+
}
}
exit(1);
}
}
-/**************************************************************************************************/
-string GetLineageCommand::removeConfidences(string tax) {
- try {
-
- string taxon = "";
- int taxLength = tax.length();
- for(int i=0;i<taxLength;i++){
- if(tax[i] == ';'){
- taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
- taxon += ";";
- }
- else{
- taxon += tax[i];
- }
- }
-
- return taxon;
- }
- catch(exception& e) {
- m->errorOut(e, "GetLineageCommand", "removeConfidences");
- 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 GetLineageCommand::readAlign(){
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(alignfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(alignfile));
+ variables["[extension]"] = m->getExtension(alignfile);
+ string outputFileName = getOutputFileName("alignreport", variables);
+
ofstream out;
m->openOutputFile(outputFileName, out);
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.close();
out.close();
- if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains does not contain any sequences from " + taxons + "."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
return 0;
projects / mothur.git / blobdiff