int MergeGroupsCommand::process(SharedRAbundVectors*& thisLookUp, ofstream& out, bool& printHeaders){
try {
vector<string> setNames = designMap->getCategory();
//create sharedRabundVectors
vector<SharedRAbundVector*> data = thisLookUp->getSharedRAbundVectors();
//create SharedRAbundVectors for the merged groups. Fill with blank rabundFloatVectors
SharedRAbundVectors* merged; merged = new SharedRAbundVectors();
for (int i = 0; i < setNames.size(); i++) {
SharedRAbundVector* myLookup = new SharedRAbundVector(thisLookUp->getNumBins());
myLookup->setLabel(thisLookUp->getLabel());
myLookup->setGroup(setNames[i]);
merged->push_back(myLookup);
}
//for each OTU
for (int j = 0; j < data[0]->getNumBins(); j++) {
if (m->getControl_pressed()) { break; }
map<string, vector<int> > otusGroupAbunds;
map<string, vector<int> >::iterator itAbunds;
//for each sample
for (int i = 0; i < data.size(); i++) {
string grouping = designMap->get(data[i]->getGroup()); //what set to your belong to
if (grouping == "not found") { m->mothurOut("[ERROR]: " + data[i]->getGroup() + " is not in your design file. Ignoring!"); m->mothurOutEndLine(); grouping = "NOTFOUND"; }
else {
//Add this OTUs values to sets abunds
itAbunds = otusGroupAbunds.find(grouping);
if (itAbunds == otusGroupAbunds.end()) { //new group
vector<int> temp;
temp.push_back(data[i]->get(j));
otusGroupAbunds[grouping] = temp;
}else {
(itAbunds->second).push_back(data[i]->get(j));
}
}
}
//find results for this bin. Set merged value for this bin in the results
for (itAbunds = otusGroupAbunds.begin(); itAbunds != otusGroupAbunds.end(); itAbunds++) {
int abund = mergeAbund(itAbunds->second);
merged->set(j, abund, itAbunds->first);
}
}
//free memory
for (int i = 0; i < data.size(); i++) { delete data[i]; }
if (m->getControl_pressed()) { delete merged; return 0; }
merged->eliminateZeroOTUS(); // remove any zero OTUs created by median option.
//print new file
merged->print(out, printHeaders);
delete merged;
return 0;
}
catch(exception& e) {
m->errorOut(e, "MergeGroupsCommand", "process");
exit(1);
}
}
int MergeGroupsCommand::processCountFile(DesignMap*& designMap){
try {
CountTable countTable;
if (!countTable.testGroups(countfile)) { m->mothurOut("[ERROR]: your countfile contains no group information, please correct.\n"); m->setControl_pressed(true); return 0; }
//read countTable
countTable.readTable(countfile, true, false);
//fill Groups - checks for "all" and for any typo groups
vector<string> nameGroups = countTable.getNamesOfGroups();
if (Groups.size() == 0) { Groups = nameGroups; }
vector<string> dnamesGroups = designMap->getNamesGroups();
//sanity check
bool error = false;
if (nameGroups.size() == dnamesGroups.size()) { //at least there are the same number
//is every group in counttable also in designmap
for (int i = 0; i < nameGroups.size(); i++) {
if (m->getControl_pressed()) { break; }
if (!util.inUsersGroups(nameGroups[i], dnamesGroups)) { error = true; break; }
}
}
if (error) { m->mothurOut("[ERROR]: Your countfile does not contain the same groups as your design file, please correct\n"); m->setControl_pressed(true); return 0; }
//user selected groups - remove some groups from table
if (Groups.size() != nameGroups.size()) {
for (int i = 0; i < nameGroups.size(); i++) {
if (!util.inUsersGroups(nameGroups[i], Groups)) { countTable.removeGroup(nameGroups[i]); }
}
}
//ask again in case order changed
nameGroups = countTable.getNamesOfGroups();
int numGroups = nameGroups.size();
//create new table
CountTable newTable;
vector<string> treatments = designMap->getCategory();
map<string, vector<int> > clearedMap;
for (int i = 0; i < treatments.size(); i++) {
newTable.addGroup(treatments[i]);
vector<int> temp;
clearedMap[treatments[i]] = temp;
}
treatments = newTable.getNamesOfGroups();
set<string> namesToRemove;
vector<string> namesOfSeqs = countTable.getNamesOfSeqs();
for (int i = 0; i < namesOfSeqs.size(); i++) {
if (m->getControl_pressed()) { break; }
vector<int> thisSeqsCounts = countTable.getGroupCounts(namesOfSeqs[i]);
map<string, vector<int> > thisSeqsMap = clearedMap;
for (int j = 0; j < numGroups; j++) {
thisSeqsMap[designMap->get(nameGroups[j])].push_back(thisSeqsCounts[j]);
}
//create new counts for seq for new table
vector<int> newCounts; int totalAbund = 0;
for (int j = 0; j < treatments.size(); j++){
int abund = mergeAbund(thisSeqsMap[treatments[j]]);
newCounts.push_back(abund); //order matters, add in count for each treatment in new table.
totalAbund += abund;
}
//add seq to new table
if(totalAbund == 0) {
namesToRemove.insert(namesOfSeqs[i]);
}else { newTable.push_back(namesOfSeqs[i], newCounts); }
}
if (error) { m->setControl_pressed(true); return 0; }
//remove sequences zeroed out by median method
if (namesToRemove.size() != 0) {
//print names
ofstream out;
string accnosFile = "accnosFile.temp";
util.openOutputFile(accnosFile, out);
//output to .accnos file
for (set<string>::iterator it = namesToRemove.begin(); it != namesToRemove.end(); it++) {
if (m->getControl_pressed()) { out.close(); util.mothurRemove(accnosFile); return 0; }
out << *it << endl;
}
out.close();
//run remove.seqs
string inputString = "accnos=" + accnosFile + ", fasta=" + fastafile;
m->mothurOut("/******************************************/"); m->mothurOutEndLine();
m->mothurOut("Running command: remove.seqs(" + inputString + ")"); m->mothurOutEndLine();
current->setMothurCalling(true);
Command* removeCommand = new RemoveSeqsCommand(inputString);
removeCommand->execute();
map<string, vector<string> > filenames = removeCommand->getOutputFiles();
delete removeCommand;
current->setMothurCalling(false);
m->mothurOut("/******************************************/"); m->mothurOutEndLine();
util.mothurRemove(accnosFile);
}
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += util.hasPath(countfile); }
map<string, string> variables;
variables["[filename]"] = thisOutputDir + util.getRootName(util.getSimpleName(countfile));
variables["[extension]"] = util.getExtension(countfile);
string outputFileName = getOutputFileName("count", variables);
outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
newTable.printTable(outputFileName);
return 0;
}
catch(exception& e) {
m->errorOut(e, "MergeGroupsCommand", "processCountFile");
exit(1);
}
}
int MergeGroupsCommand::process(vector<SharedRAbundVector*>& thisLookUp, ofstream& out){
try {
vector<string> setNames = designMap->getCategory();
if (method == "average") {
//create sharedRabundFloatVectors
vector<SharedRAbundFloatVector*> temp = thisLookUp[0]->getSharedRAbundFloatVectors(thisLookUp);
//follow code below
map<string, SharedRAbundFloatVector> merged;
map<string, SharedRAbundFloatVector>::iterator it;
for (int i = 0; i < setNames.size(); i++) {
SharedRAbundFloatVector myLookup(thisLookUp[0]->getNumBins());
myLookup.setLabel(thisLookUp[0]->getLabel());
merged[setNames[i]] = myLookup;
}
map<string, vector<int> > clearGroupAbunds;
map<string, vector<int> >::iterator itAbunds;
for (int i = 0; i < temp.size(); i++) {
if (m->control_pressed) { return 0; }
//what grouping does this group belong to
string grouping = designMap->get(temp[i]->getGroup());
if (grouping == "not found") { m->mothurOut("[ERROR]: " + temp[i]->getGroup() + " is not in your design file. Ignoring!"); m->mothurOutEndLine(); grouping = "NOTFOUND"; }
else {
//do we already have a member of this grouping?
it = merged.find(grouping);
if (it == merged.end()) { //nope, so create it
merged[grouping] = *temp[i];
merged[grouping].setGroup(grouping);
vector<int> temp;
clearGroupAbunds[grouping] = temp;
}
}
}
for (int j = 0; j < temp[0]->getNumBins(); j++) {
if (m->control_pressed) { return 0; }
map<string, vector<int> > otusGroupAbunds = clearGroupAbunds;
for (int i = 0; i < temp.size(); i++) {
string grouping = designMap->get(temp[i]->getGroup());
if (grouping == "not found") { m->mothurOut("[ERROR]: " + temp[i]->getGroup() + " is not in your design file. Ignoring!"); m->mothurOutEndLine(); grouping = "NOTFOUND"; }
else {
otusGroupAbunds[grouping].push_back(temp[i]->getAbundance(j));
}
}
for (itAbunds = otusGroupAbunds.begin(); itAbunds != otusGroupAbunds.end(); itAbunds++) {
int abund = mergeAbund(itAbunds->second);
merged[itAbunds->first].set(j, abund, itAbunds->first);
}
}
if (method == "median") {
vector<SharedRAbundFloatVector*> temp2;
for (it = merged.begin(); it != merged.end(); it++) { temp2.push_back(&(it->second)); }
eliminateZeroOTUS(temp2);
}
//print new file
for (it = merged.begin(); it != merged.end(); it++) {
if (!m->printedSharedHeaders) { (it->second).printHeaders(out); }
out << (it->second).getLabel() << '\t' << it->first << '\t';
(it->second).print(out);
}
}else {
map<string, SharedRAbundVector> merged;
map<string, SharedRAbundVector>::iterator it;
for (int i = 0; i < setNames.size(); i++) {
SharedRAbundVector myLookup(thisLookUp[0]->getNumBins());
myLookup.setLabel(thisLookUp[0]->getLabel());
merged[setNames[i]] = myLookup;
}
map<string, vector<int> > clearGroupAbunds;
map<string, vector<int> >::iterator itAbunds;
for (int i = 0; i < thisLookUp.size(); i++) {
if (m->control_pressed) { return 0; }
//what grouping does this group belong to
string grouping = designMap->get(thisLookUp[i]->getGroup());
if (grouping == "not found") { m->mothurOut("[ERROR]: " + thisLookUp[i]->getGroup() + " is not in your design file. Ignoring!"); m->mothurOutEndLine(); grouping = "NOTFOUND"; }
else {
//do we already have a member of this grouping?
it = merged.find(grouping);
if (it == merged.end()) { //nope, so create it
merged[grouping] = *thisLookUp[i];
merged[grouping].setGroup(grouping);
vector<int> temp;
clearGroupAbunds[grouping] = temp;
}
}
}
for (int j = 0; j < thisLookUp[0]->getNumBins(); j++) {
if (m->control_pressed) { return 0; }
map<string, vector<int> > otusGroupAbunds = clearGroupAbunds;
for (int i = 0; i < thisLookUp.size(); i++) {
string grouping = designMap->get(thisLookUp[i]->getGroup());
if (grouping == "not found") { m->mothurOut("[ERROR]: " + thisLookUp[i]->getGroup() + " is not in your design file. Ignoring!"); m->mothurOutEndLine(); grouping = "NOTFOUND"; }
else {
otusGroupAbunds[grouping].push_back(thisLookUp[i]->getAbundance(j));
}
}
for (itAbunds = otusGroupAbunds.begin(); itAbunds != otusGroupAbunds.end(); itAbunds++) {
int abund = mergeAbund(itAbunds->second);
merged[itAbunds->first].set(j, abund, itAbunds->first);
}
}
if (method == "median") {
vector<SharedRAbundVector*> temp;
for (it = merged.begin(); it != merged.end(); it++) { temp.push_back(&(it->second)); }
eliminateZeroOTUS(temp);
}
//print new file
for (it = merged.begin(); it != merged.end(); it++) {
if (!m->printedSharedHeaders) { (it->second).printHeaders(out); }
out << (it->second).getLabel() << '\t' << it->first << '\t';
(it->second).print(out);
}
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "MergeGroupsCommand", "process");
exit(1);
}
}
