int SensSpecCommand::processColumn(){
try{
string origCutoff = "";
bool getCutoff = 0;
if(cutoff == -1.00) { getCutoff = 1; }
else { origCutoff = toString(cutoff); cutoff += (0.49 / double(precision)); }
set<string> seqPairSet;
int numSeqs = 0;
InputData input(listFile, "list");
ListVector* list = input.getListVector();
string lastLabel = list->getLabel();
//if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
set<string> processedLabels;
set<string> userLabels = labels;
while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete list; return 0; }
if(allLines == 1 || labels.count(list->getLabel()) == 1){
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
//process
numSeqs = fillSeqPairSet(seqPairSet, list);
process(seqPairSet, list->getLabel(), getCutoff, origCutoff, numSeqs);
}
if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
string saveLabel = list->getLabel();
delete list;
list = input.getListVector(lastLabel);
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
//process
numSeqs = fillSeqPairSet(seqPairSet, list);
process(seqPairSet, list->getLabel(), getCutoff, origCutoff, numSeqs);
//restore real lastlabel to save below
list->setLabel(saveLabel);
}
lastLabel = list->getLabel();
delete list;
list = input.getListVector();
}
//output error messages about any remaining user labels
set<string>::iterator it;
bool needToRun = false;
for (it = userLabels.begin(); it != userLabels.end(); it++) {
m->mothurOut("Your file does not include the label " + *it);
if (processedLabels.count(lastLabel) != 1) {
m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
needToRun = true;
}else {
m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
}
}
//run last label if you need to
if (needToRun == true) {
if (list != NULL) { delete list; }
list = input.getListVector(lastLabel);
//process
numSeqs = fillSeqPairSet(seqPairSet, list);
delete list;
process(seqPairSet, list->getLabel(), getCutoff, origCutoff, numSeqs);
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "SensSpecCommand", "processColumn");
exit(1);
}
}
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 GetRAbundCommand::processList(ofstream& out){
try {
CountTable ct;
ct.readTable(countfile, false, false);
InputData input(inputfile, format);
ListVector* list = input.getListVector();
string lastLabel = list->getLabel();
//if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
set<string> processedLabels;
set<string> userLabels = labels;
if (m->control_pressed) { delete list; return 0; }
while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
if(allLines == 1 || labels.count(list->getLabel()) == 1){
m->mothurOut(list->getLabel()); m->mothurOutEndLine();
if (m->control_pressed) { delete list; return 0; }
RAbundVector* rabund = new RAbundVector();
createRabund(ct, list, rabund);
if(sorted) { rabund->print(out); }
else { rabund->nonSortedPrint(out); }
delete rabund;
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
}
if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
string saveLabel = list->getLabel();
delete list;
list = input.getListVector(lastLabel);
m->mothurOut(list->getLabel()); m->mothurOutEndLine();
if (m->control_pressed) { delete list; return 0; }
RAbundVector* rabund = new RAbundVector();
createRabund(ct, list, rabund);
if(sorted) { rabund->print(out); }
else { rabund->nonSortedPrint(out); }
delete rabund;
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
//restore real lastlabel to save below
list->setLabel(saveLabel);
}
lastLabel = list->getLabel();
delete list;
list = input.getListVector();
}
//output error messages about any remaining user labels
set<string>::iterator it;
bool needToRun = false;
for (it = userLabels.begin(); it != userLabels.end(); it++) {
m->mothurOut("Your file does not include the label " + *it);
if (processedLabels.count(lastLabel) != 1) {
m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
needToRun = true;
}else {
m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
}
}
//run last label if you need to
if (needToRun == true) {
if (list != NULL) { delete list; }
list = input.getListVector(lastLabel);
m->mothurOut(list->getLabel()); m->mothurOutEndLine();
if (m->control_pressed) { delete list; return 0; }
RAbundVector* rabund = new RAbundVector();
createRabund(ct, list, rabund);
if(sorted) { rabund->print(out); }
else { rabund->nonSortedPrint(out); }
delete rabund;
delete list;
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "GetRAbundCommand", "processList");
exit(1);
}
}
//**********************************************************************************************************************
//returns a vector of listVectors where "little" vector is first
vector<ListVector> OtuHierarchyCommand::getListVectors() {
try {
int pos; //to use in smart distancing, position of last read in file
int lastPos;
vector<ListVector> lists;
//if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
set<string> processedLabels;
set<string> userLabels = labels;
//open file
ifstream in;
m->openInputFile(listFile, in);
//get first list vector in file
ListVector* list = NULL;
string lastLabel = "";
if (!in.eof()) {
pos = in.tellg();
lastPos = pos;
list = new ListVector(in);
m->gobble(in);
lastLabel = list->getLabel();
}
while ((list != NULL) && (userLabels.size() != 0)) {
if (m->control_pressed) { in.close(); delete list; return lists; }
//is this a listvector that we want?
if(labels.count(list->getLabel()) == 1){
//make copy of listvector
ListVector temp(*list);
lists.push_back(temp);
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
}
//you have a label the user want that is smaller than this label and the last label has not already been processed
if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
string saveLabel = list->getLabel();
int savePos = in.tellg();
//get smart distance line
delete list;
in.seekg(lastPos);
if (!in.eof()) {
list = new ListVector(in);
}else { list = NULL; }
//make copy of listvector
ListVector temp(*list);
lists.push_back(temp);
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
//restore real lastlabel to save below
list->setLabel(saveLabel);
in.seekg(savePos);
}
lastLabel = list->getLabel();
lastPos = pos;
//get next line
delete list;
if (!in.eof()) {
pos = in.tellg();
list = new ListVector(in);
m->gobble(in);
}else { list = NULL; }
}
if (m->control_pressed) { in.close(); return lists; }
//output error messages about any remaining user labels
set<string>::iterator it;
bool needToRun = false;
for (it = userLabels.begin(); it != userLabels.end(); it++) {
m->mothurOut("Your file does not include the label " + *it);
if (processedLabels.count(lastLabel) != 1) {
m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
needToRun = true;
}else {
m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
}
}
if (m->control_pressed) { in.close(); return lists; }
//run last label if you need to
if (needToRun == true) {
if (list != NULL) { delete list; }
in.seekg(lastPos);
if (!in.eof()) {
list = new ListVector(in);
//make copy of listvector
ListVector temp(*list);
lists.push_back(temp);
delete list;
}
}
in.close();
return lists;
}
catch(exception& e) {
m->errorOut(e, "OtuHierarchyCommand", "getListVectors");
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 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 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 GetOtusCommand::readListGroup(){
try {
InputData* input = new InputData(listfile, "list");
ListVector* list = input->getListVector();
string lastLabel = list->getLabel();
//using first label seen if none is provided
if (label == "") { label = lastLabel; }
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
map<string, string> variables;
variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
variables["[tag]"] = label;
variables["[extension]"] = m->getExtension(listfile);
string outputFileName = getOutputFileName("list", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
string GroupOutputDir = outputDir;
if (outputDir == "") { GroupOutputDir += m->hasPath(groupfile); }
variables["[filename]"] = GroupOutputDir + m->getRootName(m->getSimpleName(groupfile));
variables["[extension]"] = m->getExtension(groupfile);
string outputGroupFileName = getOutputFileName("group", variables);
ofstream outGroup;
m->openOutputFile(outputGroupFileName, outGroup);
//if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
set<string> labels; labels.insert(label);
set<string> processedLabels;
set<string> userLabels = labels;
bool wroteSomething = false;
//as long as you are not at the end of the file or done wih the lines you want
while((list != NULL) && (userLabels.size() != 0)) {
if (m->control_pressed) { delete list; delete input; out.close(); outGroup.close(); m->mothurRemove(outputFileName); m->mothurRemove(outputGroupFileName);return 0; }
if(labels.count(list->getLabel()) == 1){
processList(list, groupMap, out, outGroup, wroteSomething);
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
}
if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
string saveLabel = list->getLabel();
delete list;
list = input->getListVector(lastLabel);
processList(list, groupMap, out, outGroup, wroteSomething);
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
//restore real lastlabel to save below
list->setLabel(saveLabel);
}
lastLabel = list->getLabel();
delete list; list = NULL;
//get next line to process
list = input->getListVector();
}
if (m->control_pressed) { if (list != NULL) { delete list; } delete input; out.close(); outGroup.close(); m->mothurRemove(outputFileName); m->mothurRemove(outputGroupFileName); return 0; }
//output error messages about any remaining user labels
set<string>::iterator it;
bool needToRun = false;
for (it = userLabels.begin(); it != userLabels.end(); it++) {
m->mothurOut("Your file does not include the label " + *it);
if (processedLabels.count(lastLabel) != 1) {
m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
needToRun = true;
}else {
m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
}
}
//run last label if you need to
if (needToRun == true) {
if (list != NULL) { delete list; }
list = input->getListVector(lastLabel);
processList(list, groupMap, out, outGroup, wroteSomething);
delete list; list = NULL;
}
out.close();
outGroup.close();
if (wroteSomething == false) { m->mothurOut("At distance " + label + " your file does NOT contain any otus containing sequences from the groups you wish to get."); m->mothurOutEndLine(); }
outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName);
return 0;
}
catch(exception& e) {
m->errorOut(e, "GetOtusCommand", "readList");
exit(1);
}
}
//**********************************************************************************************************************
int MGClusterCommand::execute(){
try {
if (abort == true) { if (calledHelp) { return 0; } return 2; }
//read names file
if (namefile != "") {
nameMap = new NameAssignment(namefile);
nameMap->readMap();
}else{ nameMap= new NameAssignment(); }
string fileroot = outputDir + m->getRootName(m->getSimpleName(blastfile));
string tag = "";
time_t start;
float previousDist = 0.00000;
float rndPreviousDist = 0.00000;
//read blastfile - creates sparsematrices for the distances and overlaps as well as a listvector
//must remember to delete those objects here since readBlast does not
read = new ReadBlast(blastfile, cutoff, penalty, length, minWanted, hclusterWanted);
read->read(nameMap);
list = new ListVector(nameMap->getListVector());
RAbundVector* rabund = new RAbundVector(list->getRAbundVector());
if (m->control_pressed) { outputTypes.clear(); delete nameMap; delete read; delete list; delete rabund; return 0; }
start = time(NULL);
oldList = *list;
map<string, int> Seq2Bin;
map<string, int> oldSeq2Bin;
if (method == "furthest") { tag = "fn"; }
else if (method == "nearest") { tag = "nn"; }
else { tag = "an"; }
//open output files
m->openOutputFile(fileroot+ tag + ".list", listFile);
m->openOutputFile(fileroot+ tag + ".rabund", rabundFile);
m->openOutputFile(fileroot+ tag + ".sabund", sabundFile);
if (m->control_pressed) {
delete nameMap; delete read; delete list; delete rabund;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
double saveCutoff = cutoff;
if (!hclusterWanted) {
//get distmatrix and overlap
SparseMatrix* distMatrix = read->getDistMatrix();
overlapMatrix = read->getOverlapMatrix(); //already sorted by read
delete read;
//create cluster
if (method == "furthest") { cluster = new CompleteLinkage(rabund, list, distMatrix, cutoff, method); }
else if(method == "nearest"){ cluster = new SingleLinkage(rabund, list, distMatrix, cutoff, method); }
else if(method == "average"){ cluster = new AverageLinkage(rabund, list, distMatrix, cutoff, method); }
cluster->setMapWanted(true);
Seq2Bin = cluster->getSeqtoBin();
oldSeq2Bin = Seq2Bin;
if (m->control_pressed) {
delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
//cluster using cluster classes
while (distMatrix->getSmallDist() < cutoff && distMatrix->getNNodes() > 0){
cluster->update(cutoff);
if (m->control_pressed) {
delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
float dist = distMatrix->getSmallDist();
float rndDist;
if (hard) {
rndDist = m->ceilDist(dist, precision);
}else{
rndDist = m->roundDist(dist, precision);
}
if(previousDist <= 0.0000 && dist != previousDist){
oldList.setLabel("unique");
printData(&oldList);
}
else if(rndDist != rndPreviousDist){
if (merge) {
ListVector* temp = mergeOPFs(oldSeq2Bin, rndPreviousDist);
if (m->control_pressed) {
delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster; delete temp;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
temp->setLabel(toString(rndPreviousDist, precisionLength-1));
printData(temp);
delete temp;
}else{
oldList.setLabel(toString(rndPreviousDist, precisionLength-1));
printData(&oldList);
}
}
previousDist = dist;
rndPreviousDist = rndDist;
oldList = *list;
Seq2Bin = cluster->getSeqtoBin();
oldSeq2Bin = Seq2Bin;
}
if(previousDist <= 0.0000){
oldList.setLabel("unique");
printData(&oldList);
}
else if(rndPreviousDist<cutoff){
if (merge) {
ListVector* temp = mergeOPFs(oldSeq2Bin, rndPreviousDist);
if (m->control_pressed) {
delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster; delete temp;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
temp->setLabel(toString(rndPreviousDist, precisionLength-1));
printData(temp);
delete temp;
}else{
oldList.setLabel(toString(rndPreviousDist, precisionLength-1));
printData(&oldList);
}
}
//free memory
overlapMatrix.clear();
delete distMatrix;
delete cluster;
}else { //use hcluster to cluster
//get distmatrix and overlap
overlapFile = read->getOverlapFile();
distFile = read->getDistFile();
delete read;
//sort the distance and overlap files
sortHclusterFiles(distFile, overlapFile);
if (m->control_pressed) {
delete nameMap; delete list; delete rabund;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
//create cluster
hcluster = new HCluster(rabund, list, method, distFile, nameMap, cutoff);
hcluster->setMapWanted(true);
Seq2Bin = cluster->getSeqtoBin();
oldSeq2Bin = Seq2Bin;
vector<seqDist> seqs; seqs.resize(1); // to start loop
//ifstream inHcluster;
//m->openInputFile(distFile, inHcluster);
if (m->control_pressed) {
delete nameMap; delete list; delete rabund; delete hcluster;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
while (seqs.size() != 0){
seqs = hcluster->getSeqs();
//to account for cutoff change in average neighbor
if (seqs.size() != 0) {
if (seqs[0].dist > cutoff) { break; }
}
if (m->control_pressed) {
delete nameMap; delete list; delete rabund; delete hcluster;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
m->mothurRemove(distFile);
m->mothurRemove(overlapFile);
outputTypes.clear();
return 0;
}
for (int i = 0; i < seqs.size(); i++) { //-1 means skip me
if (seqs[i].seq1 != seqs[i].seq2) {
cutoff = hcluster->update(seqs[i].seq1, seqs[i].seq2, seqs[i].dist);
if (m->control_pressed) {
delete nameMap; delete list; delete rabund; delete hcluster;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
m->mothurRemove(distFile);
m->mothurRemove(overlapFile);
outputTypes.clear();
return 0;
}
float rndDist;
if (hard) {
rndDist = m->ceilDist(seqs[i].dist, precision);
}else{
rndDist = m->roundDist(seqs[i].dist, precision);
}
if((previousDist <= 0.0000) && (seqs[i].dist != previousDist)){
oldList.setLabel("unique");
printData(&oldList);
}
else if((rndDist != rndPreviousDist)){
if (merge) {
ListVector* temp = mergeOPFs(oldSeq2Bin, rndPreviousDist);
if (m->control_pressed) {
delete nameMap; delete list; delete rabund; delete hcluster; delete temp;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
m->mothurRemove(distFile);
m->mothurRemove(overlapFile);
outputTypes.clear();
return 0;
}
temp->setLabel(toString(rndPreviousDist, precisionLength-1));
printData(temp);
delete temp;
}else{
oldList.setLabel(toString(rndPreviousDist, precisionLength-1));
printData(&oldList);
}
}
previousDist = seqs[i].dist;
rndPreviousDist = rndDist;
oldList = *list;
Seq2Bin = cluster->getSeqtoBin();
oldSeq2Bin = Seq2Bin;
}
}
}
//inHcluster.close();
if(previousDist <= 0.0000){
oldList.setLabel("unique");
printData(&oldList);
}
else if(rndPreviousDist<cutoff){
if (merge) {
ListVector* temp = mergeOPFs(oldSeq2Bin, rndPreviousDist);
if (m->control_pressed) {
delete nameMap; delete list; delete rabund; delete hcluster; delete temp;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
m->mothurRemove(distFile);
m->mothurRemove(overlapFile);
outputTypes.clear();
return 0;
}
temp->setLabel(toString(rndPreviousDist, precisionLength-1));
printData(temp);
delete temp;
}else{
oldList.setLabel(toString(rndPreviousDist, precisionLength-1));
printData(&oldList);
}
}
delete hcluster;
m->mothurRemove(distFile);
m->mothurRemove(overlapFile);
}
delete list;
delete rabund;
listFile.close();
sabundFile.close();
rabundFile.close();
if (m->control_pressed) {
delete nameMap;
listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
outputTypes.clear();
return 0;
}
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
m->mothurOut(fileroot+ tag + ".list"); m->mothurOutEndLine(); outputNames.push_back(fileroot+ tag + ".list"); outputTypes["list"].push_back(fileroot+ tag + ".list");
m->mothurOut(fileroot+ tag + ".rabund"); m->mothurOutEndLine(); outputNames.push_back(fileroot+ tag + ".rabund"); outputTypes["rabund"].push_back(fileroot+ tag + ".rabund");
m->mothurOut(fileroot+ tag + ".sabund"); m->mothurOutEndLine(); outputNames.push_back(fileroot+ tag + ".sabund"); outputTypes["sabund"].push_back(fileroot+ tag + ".sabund");
m->mothurOutEndLine();
if (saveCutoff != cutoff) {
if (hard) { saveCutoff = m->ceilDist(saveCutoff, precision); }
else { saveCutoff = m->roundDist(saveCutoff, precision); }
m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine();
}
//set list file as new current listfile
string current = "";
itTypes = outputTypes.find("list");
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
}
//set rabund file as new current rabundfile
itTypes = outputTypes.find("rabund");
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
}
//set sabund file as new current sabundfile
itTypes = outputTypes.find("sabund");
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
}
m->mothurOut("It took " + toString(time(NULL) - start) + " seconds to cluster."); m->mothurOutEndLine();
return 0;
}
catch(exception& e) {
m->errorOut(e, "MGClusterCommand", "execute");
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);
}
}