#include "getseqscommand.h"
#include "sequence.hpp"
#include "listvector.hpp"
+#include "counttable.h"
//**********************************************************************************************************************
-vector<string> GetSeqsCommand::getValidParameters(){
+vector<string> GetSeqsCommand::setParameters(){
try {
- string Array[] = {"fasta","name", "group", "alignreport", "accnos", "dups", "list","taxonomy","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ 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,false,true); parameters.push_back(ptaxonomy);
+ CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none","alignreport",false,false); parameters.push_back(palignreport);
+ CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none","qfile",false,false); parameters.push_back(pqfile);
+ CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(paccnos);
+ 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 paccnos2("accnos2", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(paccnos2);
+
+ 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, "GetSeqsCommand", "getValidParameters");
+ m->errorOut(e, "GetSeqsCommand", "setParameters");
exit(1);
}
}
+//**********************************************************************************************************************
+string GetSeqsCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, count, list, taxonomy, quality or alignreport file.\n";
+ helpString += "It outputs a file containing only the sequences in the .accnos file.\n";
+ helpString += "The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups. You must provide accnos unless you have a valid current accnos file, and at least one of the other parameters.\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=true. \n";
+ helpString += "The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
+ helpString += "Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetSeqsCommand", "getHelpString");
+ exit(1);
+ }
+}
+
//**********************************************************************************************************************
GetSeqsCommand::GetSeqsCommand(){
try {
- //initialize outputTypes
+ abort = true; calledHelp = true;
+ setParameters();
vector<string> tempOutNames;
outputTypes["fasta"] = tempOutNames;
outputTypes["taxonomy"] = tempOutNames;
outputTypes["group"] = tempOutNames;
outputTypes["alignreport"] = tempOutNames;
outputTypes["list"] = tempOutNames;
+ outputTypes["qfile"] = tempOutNames;
+ outputTypes["count"] = tempOutNames;
+ outputTypes["accnosreport"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
}
}
//**********************************************************************************************************************
-vector<string> GetSeqsCommand::getRequiredParameters(){
- try {
- string Array[] = {"accnos"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "GetSeqsCommand", "getRequiredParameters");
- exit(1);
- }
-}
-//**********************************************************************************************************************
-vector<string> GetSeqsCommand::getRequiredFiles(){
- try {
- vector<string> myArray;
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "GetSeqsCommand", "getRequiredFiles");
- exit(1);
- }
+string GetSeqsCommand::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],[distance],pick,[extension]"; }
+ else if (type == "qfile") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "accnosreport") { pattern = "[filename],pick.accnos.report"; }
+ 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, "GetSeqsCommand", "getOutputPattern");
+ exit(1);
+ }
}
//**********************************************************************************************************************
GetSeqsCommand::GetSeqsCommand(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", "group", "alignreport", "accnos", "dups", "list","taxonomy","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
outputTypes["group"] = tempOutNames;
outputTypes["alignreport"] = tempOutNames;
outputTypes["list"] = tempOutNames;
+ outputTypes["qfile"] = tempOutNames;
+ outputTypes["accnosreport"] = 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 (path == "") { parameters["accnos"] = inputDir + it->second; }
}
+ it = parameters.find("accnos2");
+ //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["accnos2"] = inputDir + it->second; }
+ }
+
it = parameters.find("list");
//user has given a template file
if(it != parameters.end()){
//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; }
+ }
+
+ 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
accnosfile = validParameter.validFile(parameters, "accnos", true);
if (accnosfile == "not open") { abort = true; }
- else if (accnosfile == "not found") { accnosfile = ""; m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
+ else if (accnosfile == "not found") {
+ accnosfile = m->getAccnosFile();
+ if (accnosfile != "") { m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
+ abort = true;
+ }
+ }else { m->setAccnosFile(accnosfile); }
+
+ if (accnosfile2 == "not found") { accnosfile2 = ""; }
fastafile = validParameter.validFile(parameters, "fasta", true);
- if (fastafile == "not open") { abort = true; }
- else if (fastafile == "not found") { fastafile = ""; }
+ 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 { 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; }
+ if (taxfile == "not open") { taxfile = ""; 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); }
+
+ accnosfile2 = validParameter.validFile(parameters, "accnos2", true);
+ if (accnosfile2 == "not open") { abort = true; }
+ else if (accnosfile2 == "not found") { accnosfile2 = ""; }
+
+ 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;
+ }
+
string usedDups = "true";
- string temp = validParameter.validFile(parameters, "dups", false); if (temp == "not found") { temp = "false"; usedDups = ""; }
+ string temp = validParameter.validFile(parameters, "dups", false); if (temp == "not found") { temp = "true"; usedDups = ""; }
dups = m->isTrue(temp);
- 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 ((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 == "") && (accnosfile2 == "") && (countfile == "")) { m->mothurOut("You must provide one of the following: fasta, name, group, count, alignreport, taxonomy, quality 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);
+ }
+ }
}
}
}
//**********************************************************************************************************************
-void GetSeqsCommand::help(){
- try {
- m->mothurOut("The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, list, taxonomy or alignreport file.\n");
- m->mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
- m->mothurOut("The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, alignreport and dups. You must provide accnos and at least one of the other parameters.\n");
- m->mothurOut("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");
- m->mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
- m->mothurOut("Example get.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");
- }
- catch(exception& e) {
- m->errorOut(e, "GetSeqsCommand", "help");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
int GetSeqsCommand::execute(){
try {
- if (abort == true) { return 0; }
+ 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 (fastafile != "") { readFasta(); }
- if (groupfile != "") { readGroup(); }
- if (alignfile != "") { readAlign(); }
- if (listfile != "") { readList(); }
- if (taxfile != "") { readTax(); }
+ if (namefile != "") { readName(); }
+ if (fastafile != "") { readFasta(); }
+ if (groupfile != "") { readGroup(); }
+ if (countfile != "") { readCount(); }
+ if (alignfile != "") { readAlign(); }
+ if (listfile != "") { readList(); }
+ if (taxfile != "") { readTax(); }
+ if (qualfile != "") { readQual(); }
+ if (accnosfile2 != "") { compareAccnos(); }
+
+ if (m->debug) { runSanityCheck(); }
+
+ if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
- if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
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); }
+ }
+
+ 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);
string name;
bool wroteSomething = false;
+ int selectedCount = 0;
+
+ if (m->debug) { set<string> temp; sanity["fasta"] = temp; }
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();
+
+ if (!dups) {//adjust name if needed
+ map<string, string>::iterator it = uniqueMap.find(name);
+ if (it != uniqueMap.end()) { currSeq.setName(it->second); }
+ }
+ name = currSeq.getName();
+
if (name != "") {
//if this name is in the accnos file
if (names.count(name) != 0) {
wroteSomething = true;
currSeq.printSequence(out);
+ selectedCount++;
+
+ if (m->debug) { sanity["fasta"].insert(name); }
}
}
m->gobble(in);
in.close();
out.close();
+
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
+
return 0;
}
}
}
//**********************************************************************************************************************
-int GetSeqsCommand::readList(){
+int GetSeqsCommand::readQual(){
try {
string thisOutputDir = outputDir;
- if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
+ if (outputDir == "") { thisOutputDir += m->hasPath(qualfile); }
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(qualfile));
+ variables["[extension]"] = m->getExtension(qualfile);
+ string outputFileName = getOutputFileName("qfile", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(qualfile, in);
+ string name;
+
+ bool wroteSomething = false;
+ int selectedCount = 0;
+
+ if (m->debug) { set<string> temp; sanity["qual"] = temp; }
+
+ while(!in.eof()){
+ string saveName = "";
+ string name = "";
+ string scores = "";
+
+ in >> name;
+
+ if (!dups) {//adjust name if needed
+ map<string, string>::iterator it = uniqueMap.find(name);
+ if (it != uniqueMap.end()) { name = it->second; }
+ }
+
+ if (name.length() != 0) {
+ saveName = name.substr(1);
+ while (!in.eof()) {
+ char c = in.get();
+ if (c == 10 || c == 13 || c == -1){ 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;
+ selectedCount++;
+ if (m->debug) { sanity["qual"].insert(name); }
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+
+ if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName); outputTypes["qfile"].push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your quality file."); m->mothurOutEndLine();
+
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetSeqsCommand", "readQual");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int GetSeqsCommand::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;
+ int selectedCount = 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;
+ selectedCount+= thisTotal;
+ }
+ }
+ in.close();
+ out.close();
+
+ //check for groups that have been eliminated
+ CountTable ct;
+ if (ct.testGroups(outputFileName)) {
+ ct.readTable(outputFileName, true, false);
+ ct.printTable(outputFileName);
+ }
+
+ if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
+ outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your count file."); m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetSeqsCommand", "readCount");
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+int GetSeqsCommand::readList(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
+ variables["[extension]"] = m->getExtension(listfile);
+
ifstream in;
m->openInputFile(listfile, in);
bool wroteSomething = false;
+ int selectedCount = 0;
+
+ if (m->debug) { set<string> temp; sanity["list"] = temp; }
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ selectedCount = 0;
//read in list vector
ListVector list(in);
//make a new list vector
ListVector newList;
newList.setLabel(list.getLabel());
+
+ variables["[distance]"] = list.getLabel();
+ string outputFileName = getOutputFileName("list", variables);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+ outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ vector<string> binLabels = list.getLabels();
+ vector<string> newBinLabels;
+
+ if (m->control_pressed) { in.close(); out.close(); return 0; }
//for each bin
for (int i = 0; i < list.getNumBins(); i++) {
//parse out names that are in accnos file
string binnames = list.get(i);
+ vector<string> bnames;
+ m->splitAtComma(binnames, bnames);
string newNames = "";
- while (binnames.find_first_of(',') != -1) {
- string name = binnames.substr(0,binnames.find_first_of(','));
- binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
-
+ for (int i = 0; i < bnames.size(); i++) {
+ string name = bnames[i];
//if that name is in the .accnos file, add it
- if (names.count(name) != 0) { newNames += name + ","; }
+ if (names.count(name) != 0) { newNames += name + ","; selectedCount++; if (m->debug) { sanity["list"].insert(name); } }
}
- //get last name
- if (names.count(binnames) != 0) { newNames += binnames + ","; }
-
//if there are names in this bin add to new list
if (newNames != "") {
newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
- newList.push_back(newNames);
+ newList.push_back(newNames);
+ newBinLabels.push_back(binLabels[i]);
}
}
//print new listvector
if (newList.getNumBins() != 0) {
wroteSomething = true;
+ newList.setLabels(newBinLabels);
+ newList.printHeaders(out);
newList.print(out);
}
m->gobble(in);
+ out.close();
}
in.close();
- out.close();
+
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
- outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " 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);
+ 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);
string name, firstCol, secondCol;
bool wroteSomething = false;
-
+ int selectedCount = 0;
+
+ if (m->debug) { set<string> temp; sanity["name"] = temp; }
+ if (m->debug) { set<string> temp; sanity["dupname"] = temp; }
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 >> firstCol; m->gobble(in);
in >> secondCol;
string hold = "";
if (dups) { hold = 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);
- }
-
- //get name after last ,
- parsedNames.push_back(secondCol);
+ m->splitAtComma(secondCol, parsedNames);
vector<string> validSecond;
for (int i = 0; i < parsedNames.size(); i++) {
if (names.count(parsedNames[i]) != 0) {
validSecond.push_back(parsedNames[i]);
+ if (m->debug) { sanity["dupname"].insert(parsedNames[i]); }
}
}
if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
- for (int i = 0; i < parsedNames.size(); i++) { names.insert(parsedNames[i]); }
+ for (int i = 0; i < parsedNames.size(); i++) { names.insert(parsedNames[i]); if (m->debug) { sanity["dupname"].insert(parsedNames[i]); } }
out << firstCol << '\t' << hold << endl;
wroteSomething = true;
+ selectedCount += parsedNames.size();
+ if (m->debug) { sanity["name"].insert(firstCol); }
}else {
+
+ selectedCount += 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) {
//you know you have at least one valid second since first column is valid
for (int i = 0; i < validSecond.size()-1; i++) { out << validSecond[i] << ','; }
out << validSecond[validSecond.size()-1] << endl;
+
+ if (m->debug) { sanity["name"].insert(firstCol); }
//make first name in set you come to first column and then add the remaining names to second column
}else {
+
//you want part of this row
if (validSecond.size() != 0) {
wroteSomething = true;
out << validSecond[0] << '\t';
+ //we are changing the unique name in the fasta file
+ uniqueMap[firstCol] = validSecond[0];
//you know you have at least one valid second since first column is valid
for (int i = 0; i < validSecond.size()-1; i++) { out << validSecond[i] << ','; }
out << validSecond[validSecond.size()-1] << endl;
+
+ if (m->debug) { sanity["name"].insert(validSecond[0]); }
}
}
}
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
+
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);
string name, group;
bool wroteSomething = false;
+ int selectedCount = 0;
+
+ if (m->debug) { set<string> temp; sanity["group"] = temp; }
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 this name is in the accnos file
if (names.count(name) != 0) {
wroteSomething = true;
out << name << '\t' << group << endl;
+ selectedCount++;
+
+ if (m->debug) { sanity["group"].insert(name); }
}
m->gobble(in);
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["group"].push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
+
+
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);
string name, tax;
bool wroteSomething = false;
+ int selectedCount = 0;
+
+ if (m->debug) { set<string> temp; sanity["tax"] = temp; }
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 (!dups) {//adjust name if needed
+ map<string, string>::iterator it = uniqueMap.find(name);
+ if (it != uniqueMap.end()) { name = it->second; }
+ }
//if this name is in the accnos file
if (names.count(name) != 0) {
wroteSomething = true;
out << name << '\t' << tax << endl;
+ selectedCount++;
+
+ if (m->debug) { sanity["tax"].insert(name); }
}
m->gobble(in);
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
return 0;
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));
+ string outputFileName = getOutputFileName("alignreport", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
string name, junk;
bool wroteSomething = false;
+ int selectedCount = 0;
//read column headers
for (int i = 0; i < 16; 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
+
+ if (!dups) {//adjust name if needed
+ map<string, string>::iterator it = uniqueMap.find(name);
+ if (it != uniqueMap.end()) { name = it->second; }
+ }
//if this name is in the accnos file
if (names.count(name) != 0) {
wroteSomething = true;
+ selectedCount++;
out << name << '\t';
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
+
return 0;
}
}
}
//**********************************************************************************************************************
+//just looking at common mistakes.
+int GetSeqsCommand::runSanityCheck(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(fastafile); }
+ string filename = outputDir + "get.seqs.debug.report";
+
+ ofstream out;
+ m->openOutputFile(filename, out);
+
+
+ //compare fasta, name, qual and taxonomy if given to make sure they contain the same seqs
+ if (fastafile != "") {
+ if (namefile != "") { //compare with fasta
+ if (sanity["fasta"] != sanity["name"]) { //create mismatch file
+ createMisMatchFile(out, fastafile, namefile, sanity["fasta"], sanity["name"]);
+ }
+ }
+ if (qualfile != "") {
+ if (sanity["fasta"] != sanity["qual"]) { //create mismatch file
+ createMisMatchFile(out, fastafile, qualfile, sanity["fasta"], sanity["qual"]);
+ }
+ }
+ if (taxfile != "") {
+ if (sanity["fasta"] != sanity["tax"]) { //create mismatch file
+ createMisMatchFile(out, fastafile, taxfile, sanity["fasta"], sanity["tax"]);
+ }
+ }
+ }
+
+ //compare dupnames, groups and list if given to make sure they match
+ if (namefile != "") {
+ if (groupfile != "") {
+ if (sanity["dupname"] != sanity["group"]) { //create mismatch file
+ createMisMatchFile(out, namefile, groupfile, sanity["dupname"], sanity["group"]);
+ }
+ }
+ if (listfile != "") {
+ if (sanity["dupname"] != sanity["list"]) { //create mismatch file
+ createMisMatchFile(out, namefile, listfile, sanity["dupname"], sanity["list"]);
+ }
+ }
+ }else{
+
+ if ((groupfile != "") && (fastafile != "")) {
+ if (sanity["fasta"] != sanity["group"]) { //create mismatch file
+ createMisMatchFile(out, fastafile, groupfile, sanity["fasta"], sanity["group"]);
+ }
+ }
+ }
+
+ out.close();
+
+ if (m->isBlank(filename)) { m->mothurRemove(filename); }
+ else { m->mothurOut("\n[DEBUG]: " + filename + " contains the file mismatches.\n");outputNames.push_back(filename); outputTypes["debug"].push_back(filename); }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+//just looking at common mistakes.
+int GetSeqsCommand::createMisMatchFile(ofstream& out, string filename1, string filename2, set<string> set1, set<string> set2){
+ try {
+ out << "****************************************" << endl << endl;
+ out << "Names unique to " << filename1 << ":\n";
+
+ //remove names in set1 that are also in set2
+ for (set<string>::iterator it = set1.begin(); it != set1.end();) {
+ string name = *it;
+
+ if (set2.count(name) == 0) { out << name << endl; } //name unique to set1
+ else { set2.erase(name); } //you are in both so erase
+ set1.erase(it++);
+ }
+
+ out << "\nNames unique to " << filename2 << ":\n";
+ //output results
+ for (set<string>::iterator it = set2.begin(); it != set2.end(); it++) { out << *it << endl; }
+
+ out << "****************************************" << endl << endl;
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
-int GetSeqsCommand::readAccnos(){
+int GetSeqsCommand::compareAccnos(){
try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(accnosfile); }
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile));
+ string outputFileName = getOutputFileName("accnosreport", variables);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
ifstream in;
- m->openInputFile(accnosfile, in);
+ m->openInputFile(accnosfile2, in);
string name;
+ set<string> namesAccnos2;
+ set<string> namesDups;
+ set<string> namesAccnos = names;
+
+ map<string, int> nameCount;
+
+ if (namefile != "") {
+ ifstream inName;
+ m->openInputFile(namefile, inName);
+
+
+ while(!inName.eof()){
+
+ if (m->control_pressed) { inName.close(); return 0; }
+
+ string thisname, repnames;
+
+ inName >> thisname; m->gobble(inName); //read from first column
+ inName >> repnames; //read from second column
+
+ int num = m->getNumNames(repnames);
+ nameCount[thisname] = num;
+
+ m->gobble(inName);
+ }
+ inName.close();
+ }
+
while(!in.eof()){
in >> name;
-
- names.insert(name);
+
+ if (namesAccnos.count(name) == 0){ //name unique to accnos2
+ int pos = name.find_last_of('_');
+ string tempName = name;
+ if (pos != string::npos) { tempName = tempName.substr(pos+1); cout << tempName << endl; }
+ if (namesAccnos.count(tempName) == 0){
+ namesAccnos2.insert(name);
+ }else { //you are in both so erase
+ namesAccnos.erase(name);
+ namesDups.insert(name);
+ }
+ }else { //you are in both so erase
+ namesAccnos.erase(name);
+ namesDups.insert(name);
+ }
m->gobble(in);
}
in.close();
- return 0;
+ out << "Names in both files : " + toString(namesDups.size()) << endl;
+ m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
+
+ for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
+ out << (*it);
+ if (namefile != "") { out << '\t' << nameCount[(*it)]; }
+ out << endl;
+ }
+
+ out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
+ m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
+
+ for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
+ out << (*it);
+ if (namefile != "") { out << '\t' << nameCount[(*it)]; }
+ out << endl;
+ }
+
+ out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
+ m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
+
+ for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
+ out << (*it);
+ if (namefile != "") { out << '\t' << nameCount[(*it)]; }
+ out << endl;
+ }
+ out.close();
+
+ outputNames.push_back(outputFileName); outputTypes["accnosreport"].push_back(outputFileName);
+
+ return 0;
+
}
catch(exception& e) {
- m->errorOut(e, "GetSeqsCommand", "readAccnos");
+ m->errorOut(e, "GetSeqsCommand", "compareAccnos");
exit(1);
}
}
+
//**********************************************************************************************************************
projects / mothur.git / blobdiff