if(globaldata->gListVector != NULL) {
listOfNames = new ListVector(*globaldata->gListVector);
+ vector<string> names;
+ string binnames;
//map names to rows in sparsematrix
for (int i = 0; i < listOfNames->size(); i++) {
- nameToIndex[listOfNames->get(i)] = i;
+ names.clear();
+ binnames = listOfNames->get(i);
+ splitAtComma(binnames, names);
+
+ for (int j = 0; j < names.size(); j++) {
+ nameToIndex[names[j]] = i;
+ }
}
- }else { cout << "error" << endl; }
+ }else { cout << "error, no listvector." << endl; }
fastafile = globaldata->getFastaFile();
namesfile = globaldata->getNameFile();
+ groupfile = globaldata->getGroupFile();
+
+ if (groupfile != "") {
+ //read in group map info.
+ groupMap = new GroupMap(groupfile);
+ groupMap->readMap();
+ }
+
openInputFile(fastafile, in);
fasta = new FastaMap();
//**********************************************************************************************************************
GetOTURepCommand::~GetOTURepCommand(){
- delete matrix;
- delete list;
delete input;
delete read;
delete fasta;
+ if (groupfile != "") {
+ delete groupMap;
+ }
}
//**********************************************************************************************************************
int GetOTURepCommand::execute(){
try {
int count = 1;
- string nameRep, name, sequence;
+ int error;
//read fastafile
fasta->readFastaFile(in);
}
//read list file
- read = new ReadPhilFile(globaldata->getListFile());
+ read = new ReadOTUFile(globaldata->getListFile());
read->read(&*globaldata);
input = globaldata->ginput;
list = globaldata->gListVector;
+ ListVector* lastList = list;
+
+ //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 = globaldata->labels;
+ set<int> userLines = globaldata->lines;
+
- while(list != NULL){
+ while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
-
- //create output file
- string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".rep.fasta";
- openOutputFile(outputFileName, out);
-
- cout << list->getLabel() << '\t' << count << endl;
-
- //for each bin in the list vector
- for (int i = 0; i < list->size(); i++) {
- nameRep = FindRep(i);
+ cout << list->getLabel() << '\t' << count << endl;
+ error = process(list);
+ if (error == 1) { return 0; } //there is an error in hte input files, abort command
- //print out name and sequence for that bin
- sequence = fasta->getSequence(nameRep);
-
- if (sequence != "not found") {
- nameRep = nameRep + "|" + toString(i+1);
- out << ">" << nameRep << endl;
- out << sequence << endl;
- }else {
- cout << nameRep << " is missing from your fasta or name file. Please correct. " << endl;
- remove(outputFileName.c_str());
- return 0;
- }
- }
-
- out.close();
+ processedLabels.insert(list->getLabel());
+ userLabels.erase(list->getLabel());
+ userLines.erase(count);
}
+ if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
+ cout << lastList->getLabel() << '\t' << count << endl;
+ error = process(lastList);
+ if (error == 1) { return 0; } //there is an error in hte input files, abort command
+
+ processedLabels.insert(lastList->getLabel());
+ userLabels.erase(lastList->getLabel());
+ }
+
+ if (count != 1) { delete lastList; }
+ lastList = list;
+
list = input->getListVector();
count++;
}
-
+ //output error messages about any remaining user labels
+ set<string>::iterator it;
+ bool needToRun = false;
+ for (it = userLabels.begin(); it != userLabels.end(); it++) {
+ cout << "Your file does not include the label "<< *it;
+ if (processedLabels.count(lastList->getLabel()) != 1) {
+ cout << ". I will use " << lastList->getLabel() << "." << endl;
+ needToRun = true;
+ }else {
+ cout << ". Please refer to " << lastList->getLabel() << "." << endl;
+ }
+ }
+
+ //run last line if you need to
+ if (needToRun == true) {
+ cout << lastList->getLabel() << '\t' << count << endl;
+ error = process(lastList);
+ if (error == 1) { return 0; } //there is an error in hte input files, abort command
+ }
+ delete lastList;
+
+ delete matrix;
+ globaldata->gSparseMatrix = NULL;
+ delete list;
+ globaldata->gListVector = NULL;
+
return 0;
}
catch(exception& e) {
}
}
//**********************************************************************************************************************
-string GetOTURepCommand::FindRep(int bin) {
+string GetOTURepCommand::FindRep(int bin, string& group, ListVector* thisList) {
try{
vector<string> names;
map<string, float> sums;
string binnames;
float min = 10000;
string minName;
+ map<string, string> groups;
+ map<string, string>::iterator groupIt;
- binnames = list->get(bin);
-
+ binnames = thisList->get(bin);
+
//parse names into vector
splitAtComma(binnames, names);
+ //if you have a groupfile
+ if(groupfile != "") {
+ //find the groups that are in this bin
+ for (int i = 0; i < names.size(); i++) {
+ string groupName = groupMap->getGroup(names[i]);
+ if (groupName == "not found") {
+ cout << names[i] << " is missing from your group file. Please correct. " << endl;
+ groupError = true;
+ }else{
+ groups[groupName] = groupName;
+ }
+ }
+
+ //turn the groups into a string
+ for(groupIt = groups.begin(); groupIt != groups.end(); groupIt++) { group += groupIt->first + "-"; }
+
+ //rip off last dash
+ group = group.substr(0, group.length()-1);
+ }
+
//if only 1 sequence in bin then that's the rep
if (names.size() == 1) { return names[0]; }
else {
//fill binMap
for (int i = 0; i < names.size(); i++) {
for (it3 = nameToIndex.begin(); it3 != nameToIndex.end(); it3++) {
+
if (it3->first == names[i]) {
binMap[it3->second] = it3->first;
}
}
+//**********************************************************************************************************************
+int GetOTURepCommand::process(ListVector* processList) {
+ try{
+ string nameRep, name, sequence;
+
+ //create output file
+ string outputFileName = getRootName(globaldata->getListFile()) + processList->getLabel() + ".rep.fasta";
+ openOutputFile(outputFileName, out);
+
+ //for each bin in the list vector
+ for (int i = 0; i < processList->size(); i++) {
+ string groups;
+ nameRep = FindRep(i, groups, processList);
+
+ //print out name and sequence for that bin
+ sequence = fasta->getSequence(nameRep);
+
+ if (sequence != "not found") {
+ if (groupfile == "") {
+ nameRep = nameRep + "|" + toString(i+1);
+ out << ">" << nameRep << endl;
+ out << sequence << endl;
+ }else {
+ nameRep = nameRep + "|" + groups + "|" + toString(i+1);
+ out << ">" << nameRep << endl;
+ out << sequence << endl;
+ }
+ }else {
+ cout << nameRep << " is missing from your fasta or name file. Please correct. " << endl;
+ remove(outputFileName.c_str());
+ return 1;
+ }
+ }
+
+ out.close();
+ return 0;
+
+ }
+ catch(exception& e) {
+ cout << "Standard Error: " << e.what() << " has occurred in the GetOTURepCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ exit(1);
+ }
+ catch(...) {
+ cout << "An unknown error has occurred in the GetOTURepCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+
projects / mothur.git / blobdiff