string SensSpecCommand::preProcessList(){
try {
set<string> uniqueNames;
//get unique names from distance file
if (format == "phylip") {
ifstream phylipFile;
m->openInputFile(distFile, phylipFile);
string numTest;
int pNumSeqs;
phylipFile >> numTest; m->gobble(phylipFile);
if (!m->isContainingOnlyDigits(numTest)) { m->mothurOut("[ERROR]: expected a number and got " + numTest + ", quitting."); m->mothurOutEndLine(); exit(1); }
else {
m->mothurConvert(numTest, pNumSeqs);
}
string seqName;
for(int i=0;i<pNumSeqs;i++){
if (m->control_pressed) { return ""; }
phylipFile >> seqName; m->getline(phylipFile); m->gobble(phylipFile);
uniqueNames.insert(seqName);
}
phylipFile.close();
}else {
ifstream columnFile;
m->openInputFile(distFile, columnFile);
string seqNameA, seqNameB;
double distance;
while(columnFile){
if (m->control_pressed) { return ""; }
columnFile >> seqNameA >> seqNameB >> distance;
uniqueNames.insert(seqNameA); uniqueNames.insert(seqNameB);
m->gobble(columnFile);
}
columnFile.close();
}
//read list file, if numSeqs > unique names then remove redundant names
string newListFile = listFile + ".temp";
ofstream out;
m->openOutputFile(newListFile, out);
ifstream in;
m->openInputFile(listFile, in);
bool wroteSomething = false;
while(!in.eof()){
if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(newListFile); return ""; }
//read in list vector
ListVector list(in);
//listfile is already unique
if (list.getNumSeqs() == uniqueNames.size()) { in.close(); out.close(); m->mothurRemove(newListFile); return ""; }
//make a new list vector
ListVector newList;
newList.setLabel(list.getLabel());
//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 = "";
for (int j = 0; j < bnames.size(); j++) {
string name = bnames[j];
//if that name is in the .accnos file, add it
if (uniqueNames.count(name) != 0) { newNames += name + ","; }
}
//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);
}
}
//print new listvector
if (newList.getNumBins() != 0) {
wroteSomething = true;
newList.print(out);
}
m->gobble(in);
}
in.close();
out.close();
if (wroteSomething) { return newListFile; }
else { m->mothurRemove(newListFile); }
return "";
}
//**********************************************************************************************************************
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()){
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 = "";
for (int j = 0; j < bnames.size(); j++) {
string name = bnames[j];
//if that name is in the .accnos file, add it
if (names.count(name) != 0) { newNames += name + ","; selectedCount++; if (m->debug) { sanity["list"].insert(name); } }
}
//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);
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();
if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine(); }
m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
return 0;
}
catch(exception& e) {
m->errorOut(e, "GetSeqsCommand", "readList");
exit(1);
}
}
//**********************************************************************************************************************
int RemoveRareCommand::processList(){
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
string outputGroupFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
ofstream out, outGroup;
m->openOutputFile(outputFileName, out);
bool wroteSomething = false;
//you must provide a label because the names in the listfile need to be consistent
string thisLabel = "";
if (allLines) { m->mothurOut("For the listfile you must select one label, using first label in your listfile."); m->mothurOutEndLine(); }
else if (labels.size() > 1) { m->mothurOut("For the listfile you must select one label, using " + (*labels.begin()) + "."); m->mothurOutEndLine(); thisLabel = *labels.begin(); }
else { thisLabel = *labels.begin(); }
InputData input(listfile, "list");
ListVector* list = input.getListVector();
//get first one or the one we want
if (thisLabel != "") {
//use smart distancing
set<string> userLabels; userLabels.insert(thisLabel);
set<string> processedLabels;
string lastLabel = list->getLabel();
while((list != NULL) && (userLabels.size() != 0)) {
if(userLabels.count(list->getLabel()) == 1){
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
break;
}
if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
delete list;
list = input.getListVector(lastLabel);
break;
}
lastLabel = list->getLabel();
delete list;
list = input.getListVector();
}
if (userLabels.size() != 0) {
m->mothurOut("Your file does not include the label " + thisLabel + ". I will use " + lastLabel + "."); m->mothurOutEndLine();
list = input.getListVector(lastLabel);
}
}
//if groupfile is given then use it
GroupMap* groupMap;
if (groupfile != "") {
groupMap = new GroupMap(groupfile); groupMap->readMap();
SharedUtil util;
vector<string> namesGroups = groupMap->getNamesOfGroups();
util.setGroups(Groups, namesGroups);
m->openOutputFile(outputGroupFileName, outGroup);
}
if (list != NULL) {
//make a new list vector
ListVector newList;
newList.setLabel(list->getLabel());
//for each bin
for (int i = 0; i < list->getNumBins(); i++) {
if (m->control_pressed) { if (groupfile != "") { delete groupMap; outGroup.close(); m->mothurRemove(outputGroupFileName); } out.close(); m->mothurRemove(outputFileName); return 0; }
//parse out names that are in accnos file
string binnames = list->get(i);
vector<string> names;
string saveBinNames = binnames;
m->splitAtComma(binnames, names);
vector<string> newGroupFile;
if (groupfile != "") {
vector<string> newNames;
saveBinNames = "";
for(int k = 0; k < names.size(); k++) {
string group = groupMap->getGroup(names[k]);
if (m->inUsersGroups(group, Groups)) {
newGroupFile.push_back(names[k] + "\t" + group);
newNames.push_back(names[k]);
saveBinNames += names[k] + ",";
}
}
names = newNames;
saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
}
if (names.size() > nseqs) { //keep bin
newList.push_back(saveBinNames);
for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }
}
}
//print new listvector
if (newList.getNumBins() != 0) {
wroteSomething = true;
newList.print(out);
}
}
out.close();
if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
if (wroteSomething == false) { m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine(); }
outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
return 0;
}
catch(exception& e) {
m->errorOut(e, "RemoveRareCommand", "processList");
exit(1);
}
}
//**********************************************************************************************************************
int RemoveGroupsCommand::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 removedCount = 0;
while(!in.eof()){
removedCount = 0;
//read in list vector
ListVector list(in);
variables["[tag]"] = 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;
//make a new list vector
ListVector newList;
newList.setLabel(list.getLabel());
//for each bin
for (int i = 0; i < list.getNumBins(); i++) {
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);
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());
//if that name is in the .accnos file, add it
if (names.count(name) == 0) { newNames += name + ","; }
else {
//if you are not in the accnos file check if you are a name that needs to be changed
map<string, string>::iterator it = uniqueToRedundant.find(name);
if (it != uniqueToRedundant.end()) {
newNames += it->second + ",";
}else { removedCount++; }
}
}
//get last name
if (names.count(binnames) == 0) { newNames += binnames + ","; }
else { //if you are not in the accnos file check if you are a name that needs to be changed
map<string, string>::iterator it = uniqueToRedundant.find(binnames);
if (it != uniqueToRedundant.end()) {
newNames += it->second + ",";
}else { removedCount++; }
}
//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);
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();
if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine(); }
m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
return 0;
}
catch(exception& e) {
m->errorOut(e, "RemoveGroupsCommand", "readList");
exit(1);
}
}
//**********************************************************************************************************************
int GetOtusCommand::processList(ListVector*& list, GroupMap*& groupMap, ofstream& out, ofstream& outGroup, bool& wroteSomething){
try {
//make a new list vector
ListVector newList;
newList.setLabel(list->getLabel());
int numOtus = 0;
//for each bin
vector<string> binLabels = list->getLabels();
vector<string> newBinLabels;
for (int i = 0; i < list->getNumBins(); i++) {
if (m->control_pressed) { return 0; }
//parse out names that are in accnos file
string binnames = list->get(i);
bool keepBin = false;
string groupFileOutput = "";
//parse names
string individual = "";
int length = binnames.length();
for(int j=0;j<length;j++){
if(binnames[j] == ','){
string group = groupMap->getGroup(individual);
if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
if (m->inUsersGroups(group, Groups)) { keepBin = true; }
groupFileOutput += individual + "\t" + group + "\n";
individual = "";
}
else{ individual += binnames[j]; }
}
string group = groupMap->getGroup(individual);
if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
if (m->inUsersGroups(group, Groups)) { keepBin = true; }
groupFileOutput += individual + "\t" + group + "\n";
//if there are sequences from the groups we want in this bin add to new list, output to groupfile
if (keepBin) {
newList.push_back(binnames);
newBinLabels.push_back(binLabels[i]);
outGroup << groupFileOutput;
numOtus++;
}
}
//print new listvector
if (newList.getNumBins() != 0) {
wroteSomething = true;
newList.setLabels(newBinLabels);
newList.printHeaders(out);
newList.print(out);
}
m->mothurOut(newList.getLabel() + " - selected " + toString(numOtus) + " of the " + toString(list->getNumBins()) + " OTUs."); m->mothurOutEndLine();
return 0;
}
catch(exception& e) {
m->errorOut(e, "GetOtusCommand", "processList");
exit(1);
}
}
//**********************************************************************************************************************
int GetSeqsCommand::readList(){
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
ofstream out;
m->openOutputFile(outputFileName, out);
ifstream in;
m->openInputFile(listfile, in);
bool wroteSomething = false;
int selectedCount = 0;
while(!in.eof()){
selectedCount = 0;
if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
//read in list vector
ListVector list(in);
//make a new list vector
ListVector newList;
newList.setLabel(list.getLabel());
//for each bin
for (int i = 0; i < list.getNumBins(); i++) {
//parse out names that are in accnos file
string binnames = list.get(i);
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());
//if that name is in the .accnos file, add it
if (names.count(name) != 0) { newNames += name + ","; selectedCount++; }
}
//get last name
if (names.count(binnames) != 0) { newNames += binnames + ","; selectedCount++; }
//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);
}
}
//print new listvector
if (newList.getNumBins() != 0) {
wroteSomething = true;
newList.print(out);
}
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["list"].push_back(outputFileName);
m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
return 0;
}
catch(exception& e) {
m->errorOut(e, "GetSeqsCommand", "readList");
exit(1);
}
}