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);
}
}
int GetSharedOTUCommand::execute(){
try {
if (abort == true) { if (calledHelp) { return 0; } return 2; }
if ( sharedfile != "") { runShared(); }
else {
m->setGroups(Groups);
groupMap = new GroupMap(groupfile);
int error = groupMap->readMap();
if (error == 1) { delete groupMap; return 0; }
if (m->control_pressed) { delete groupMap; return 0; }
if (Groups.size() == 0) {
Groups = groupMap->getNamesOfGroups();
//make string for outputfile name
userGroups = "unique.";
for(int i = 0; i < Groups.size(); i++) { userGroups += Groups[i] + "-"; }
userGroups = userGroups.substr(0, userGroups.length()-1);
}else{
//sanity check for group names
SharedUtil util;
vector<string> namesOfGroups = groupMap->getNamesOfGroups();
util.setGroups(Groups, namesOfGroups);
groupMap->setNamesOfGroups(namesOfGroups);
}
//put groups in map to find easier
for(int i = 0; i < Groups.size(); i++) {
groupFinder[Groups[i]] = Groups[i];
}
if (fastafile != "") {
ifstream inFasta;
m->openInputFile(fastafile, inFasta);
while(!inFasta.eof()) {
if (m->control_pressed) { outputTypes.clear(); inFasta.close(); delete groupMap; return 0; }
Sequence seq(inFasta); m->gobble(inFasta);
if (seq.getName() != "") { seqs.push_back(seq); }
}
inFasta.close();
}
ListVector* lastlist = NULL;
string lastLabel = "";
//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;
ifstream in;
m->openInputFile(listfile, in);
//as long as you are not at the end of the file or done wih the lines you want
while((!in.eof()) && ((allLines == 1) || (userLabels.size() != 0))) {
if (m->control_pressed) {
if (lastlist != NULL) { delete lastlist; }
for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } outputTypes.clear();
delete groupMap; return 0;
}
list = new ListVector(in);
if(allLines == 1 || labels.count(list->getLabel()) == 1){
m->mothurOut(list->getLabel());
process(list);
processedLabels.insert(list->getLabel());
userLabels.erase(list->getLabel());
}
if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
string saveLabel = list->getLabel();
m->mothurOut(lastlist->getLabel());
process(lastlist);
processedLabels.insert(lastlist->getLabel());
userLabels.erase(lastlist->getLabel());
//restore real lastlabel to save below
list->setLabel(saveLabel);
}
lastLabel = list->getLabel();
if (lastlist != NULL) { delete lastlist; }
lastlist = list;
}
in.close();
//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) {
m->mothurOut(lastlist->getLabel());
process(lastlist);
processedLabels.insert(lastlist->getLabel());
userLabels.erase(lastlist->getLabel());
}
//reset groups parameter
m->clearGroups();
if (lastlist != NULL) { delete lastlist; }
if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete groupMap; return 0; }
}
//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); }
}
if (output == "accnos") {
itTypes = outputTypes.find("accnos");
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
}
}
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();
return 0;
}
catch(exception& e) {
m->errorOut(e, "GetSharedOTUCommand", "execute");
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 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);
}
}
//**********************************************************************************************************************
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 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);
}
}