//**********************************************************************************************************************
vector<string> RemoveRareCommand::setParameters(){
try {
- CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(plist);
- CommandParameter prabund("rabund", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(prabund);
- CommandParameter psabund("sabund", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(psabund);
- CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pshared);
- CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
- CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
- CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
- CommandParameter pnseqs("nseqs", "Number", "", "0", "", "", "",false,true); parameters.push_back(pnseqs);
- CommandParameter pbygroup("bygroup", "Boolean", "", "f", "", "", "",false,true); parameters.push_back(pbygroup);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "atleast", "none","list",false,false,true); parameters.push_back(plist);
+ CommandParameter prabund("rabund", "InputTypes", "", "", "none", "atleast", "none","rabund",false,false,true); parameters.push_back(prabund);
+ CommandParameter psabund("sabund", "InputTypes", "", "", "none", "atleast", "none","sabund",false,false,true); parameters.push_back(psabund);
+ CommandParameter pshared("shared", "InputTypes", "", "", "none", "atleast", "none","shared",false,false,true); parameters.push_back(pshared);
+ CommandParameter pcount("count", "InputTypes", "", "", "CountGroup", "none", "none","count",false,false); parameters.push_back(pcount);
+ CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none","group",false,false); parameters.push_back(pgroup);
+ CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
+ CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
+ CommandParameter pnseqs("nseqs", "Number", "", "0", "", "", "","",false,true,true); parameters.push_back(pnseqs);
+ CommandParameter pbygroup("bygroup", "Boolean", "", "f", "", "", "","",false,false); parameters.push_back(pbygroup);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
vector<string> myArray;
for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
string RemoveRareCommand::getHelpString(){
try {
string helpString = "";
- helpString += "The remove.rare command parameters are list, rabund, sabund, shared, group, label, groups, bygroup and nseqs.\n";
+ helpString += "The remove.rare command parameters are list, rabund, sabund, shared, group, count, label, groups, bygroup and nseqs.\n";
helpString += "The remove.rare command reads one of the following file types: list, rabund, sabund or shared file. It outputs a new file after removing the rare otus.\n";
helpString += "The groups parameter allows you to specify which of the groups you would like analyzed. Default=all. You may separate group names with dashes.\n";
helpString += "The label parameter is used to analyze specific labels in your input. default=all. You may separate label names with dashes.\n";
}
}
//**********************************************************************************************************************
+string RemoveRareCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "rabund") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "sabund") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "group") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "count") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "list") { pattern = "[filename],pick,[extension]"; }
+ else if (type == "shared") { pattern = "[filename],[tag],pick,[extension]"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveRareCommand", "getOutputPattern");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
RemoveRareCommand::RemoveRareCommand(){
try {
abort = true; calledHelp = true;
outputTypes["sabund"] = tempOutNames;
outputTypes["list"] = tempOutNames;
outputTypes["group"] = tempOutNames;
+ outputTypes["count"] = tempOutNames;
outputTypes["shared"] = tempOutNames;
}
catch(exception& e) {
outputTypes["list"] = tempOutNames;
outputTypes["group"] = tempOutNames;
outputTypes["shared"] = tempOutNames;
+ outputTypes["count"] = tempOutNames;
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["shared"] = inputDir + it->second; }
}
+
+ it = parameters.find("count");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["count"] = inputDir + it->second; }
+ }
}
//check for file parameters
listfile = validParameter.validFile(parameters, "list", true);
if (listfile == "not open") { abort = true; }
- else if (listfile == "not found") { listfile = ""; }
+ else if (listfile == "not found") { listfile = ""; }
+ else { m->setListFile(listfile); }
sabundfile = validParameter.validFile(parameters, "sabund", true);
if (sabundfile == "not open") { abort = true; }
else if (sabundfile == "not found") { sabundfile = ""; }
+ else { m->setSabundFile(sabundfile); }
rabundfile = validParameter.validFile(parameters, "rabund", true);
if (rabundfile == "not open") { abort = true; }
else if (rabundfile == "not found") { rabundfile = ""; }
+ else { m->setRabundFile(rabundfile); }
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { groupfile = ""; abort = true; }
else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
sharedfile = validParameter.validFile(parameters, "shared", true);
if (sharedfile == "not open") { sharedfile = ""; abort = true; }
else if (sharedfile == "not found") { sharedfile = ""; }
+ else { m->setSharedFile(sharedfile); }
+
+ countfile = validParameter.validFile(parameters, "count", true);
+ if (countfile == "not open") { countfile = ""; abort = true; }
+ else if (countfile == "not found") { countfile = ""; }
+ else { m->setCountTableFile(countfile); }
+
+ if ((groupfile != "") && (countfile != "")) {
+ m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
+ }
if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
//is there are current file available for any of these?
string temp = validParameter.validFile(parameters, "nseqs", false);
if (temp == "not found") { m->mothurOut("nseqs is a required parameter."); m->mothurOutEndLine(); abort = true; }
- else { convert(temp, nseqs); }
+ else { m->mothurConvert(temp, nseqs); }
temp = validParameter.validFile(parameters, "bygroup", false); if (temp == "not found") { temp = "f"; }
byGroup = m->isTrue(temp);
if (byGroup && (sharedfile == "")) { m->mothurOut("The byGroup parameter is only valid with a shared file."); m->mothurOutEndLine(); }
- if ((groupfile != "") && (listfile == "")) { m->mothurOut("A groupfile is only valid with a list file."); m->mothurOutEndLine(); groupfile = ""; }
+ if (((groupfile != "") || (countfile != "")) && (listfile == "")) { m->mothurOut("A group or count file is only valid with a list file."); m->mothurOutEndLine(); groupfile = ""; countfile = ""; }
}
}
if (listfile != "") { processList(); }
if (sharedfile != "") { processShared(); }
- if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
+ if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
if (outputNames.size() != 0) {
m->mothurOutEndLine();
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
}
+
+ itTypes = outputTypes.find("count");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
+ }
}
return 0;
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);
-
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
+ variables["[extension]"] = m->getExtension(listfile);
+ string outputFileName = getOutputFileName("list", variables);
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(groupfile));
+ variables["[extension]"] = m->getExtension(groupfile);
+ string outputGroupFileName = getOutputFileName("group", variables);
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
+ variables["[extension]"] = m->getExtension(countfile);
+ string outputCountFileName = getOutputFileName("count", variables);
+
ofstream out, outGroup;
m->openOutputFile(outputFileName, out);
//if groupfile is given then use it
GroupMap* groupMap;
+ CountTable ct;
if (groupfile != "") {
groupMap = new GroupMap(groupfile); groupMap->readMap();
SharedUtil util;
- util.setGroups(Groups, groupMap->namesOfGroups);
+ vector<string> namesGroups = groupMap->getNamesOfGroups();
+ util.setGroups(Groups, namesGroups);
m->openOutputFile(outputGroupFileName, outGroup);
- }
+ }else if (countfile != "") {
+ ct.readTable(countfile, true);
+ if (ct.hasGroupInfo()) {
+ vector<string> namesGroups = ct.getNamesOfGroups();
+ SharedUtil util;
+ util.setGroups(Groups, namesGroups);
+ }
+ }
if (list != NULL) {
//for each bin
for (int i = 0; i < list->getNumBins(); i++) {
- if (m->control_pressed) { if (groupfile != "") { delete groupMap; outGroup.close(); remove(outputGroupFileName.c_str()); } out.close(); remove(outputFileName.c_str()); return 0; }
+ 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);
+ int binsize = names.size();
vector<string> newGroupFile;
if (groupfile != "") {
saveBinNames += names[k] + ",";
}
}
- names = newNames;
+ names = newNames; binsize = names.size();
saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
- }
+ }else if (countfile != "") {
+ saveBinNames = "";
+ binsize = 0;
+ for(int k = 0; k < names.size(); k++) {
+ if (ct.hasGroupInfo()) {
+ vector<string> thisSeqsGroups = ct.getGroups(names[k]);
+
+ int thisSeqsCount = 0;
+ for (int n = 0; n < thisSeqsGroups.size(); n++) {
+ if (m->inUsersGroups(thisSeqsGroups[n], Groups)) {
+ thisSeqsCount += ct.getGroupCount(names[k], thisSeqsGroups[n]);
+ }
+ }
+ binsize += thisSeqsCount;
+ //if you don't have any seqs from the groups the user wants, then remove you.
+ if (thisSeqsCount == 0) { newGroupFile.push_back(names[k]); }
+ else { saveBinNames += names[k] + ","; }
+ }else {
+ binsize += ct.getNumSeqs(names[k]);
+ saveBinNames += names[k] + ",";
+ }
+ }
+ saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
+ }
- if (names.size() > nseqs) { //keep bin
+ if (binsize > nseqs) { //keep bin
newList.push_back(saveBinNames);
- for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }
- }
+ if (groupfile != "") { for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; } }
+ else if (countfile != "") { for(int k = 0; k < newGroupFile.size(); k++) { ct.remove(newGroupFile[k]); } }
+ }else { if (countfile != "") { for(int k = 0; k < names.size(); k++) { ct.remove(names[k]); } } }
}
//print new listvector
out.close();
if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
+ if (countfile != "") {
+ if (ct.hasGroupInfo()) {
+ vector<string> allGroups = ct.getNamesOfGroups();
+ for (int i = 0; i < allGroups.size(); i++) {
+ if (!m->inUsersGroups(allGroups[i], Groups)) { ct.removeGroup(allGroups[i]); }
+ }
+
+ }
+ ct.printTable(outputCountFileName);
+ outputTypes["count"].push_back(outputCountFileName); outputNames.push_back(outputCountFileName);
+ }
if (wroteSomething == false) { m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine(); }
outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(sabundfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "pick" + m->getExtension(sabundfile);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile));
+ variables["[extension]"] = m->getExtension(sabundfile);
+ string outputFileName = getOutputFileName("sabund", variables);
outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
ofstream out;
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(rabundfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "pick" + m->getExtension(rabundfile);
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile));
+ variables["[extension]"] = m->getExtension(rabundfile);
+ string outputFileName = getOutputFileName("rabund", variables);
outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
ofstream out;
//**********************************************************************************************************************
int RemoveRareCommand::processShared(){
try {
- m->Groups = Groups;
-
- string thisOutputDir = outputDir;
- if (outputDir == "") { thisOutputDir += m->hasPath(sharedfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "pick" + m->getExtension(sharedfile);
- outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
-
- ofstream out;
- m->openOutputFile(outputFileName, out);
+ m->setGroups(Groups);
//if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
InputData input(sharedfile, "sharedfile");
while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
- if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } out.close(); return 0; }
+ if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
processedLabels.insert(lookup[0]->getLabel());
userLabels.erase(lookup[0]->getLabel());
- processLookup(lookup, out);
+ processLookup(lookup);
}
if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
processedLabels.insert(lookup[0]->getLabel());
userLabels.erase(lookup[0]->getLabel());
- processLookup(lookup, out);
+ processLookup(lookup);
//restore real lastlabel to save below
lookup[0]->setLabel(saveLabel);
lookup = input.getSharedRAbundVectors();
}
- if (m->control_pressed) { out.close(); return 0; }
+ if (m->control_pressed) { return 0; }
//output error messages about any remaining user labels
set<string>::iterator it;
lookup = input.getSharedRAbundVectors(lastLabel);
m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
-
- processLookup(lookup, out);
+ processLookup(lookup);
for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
}
}
}
//**********************************************************************************************************************
-int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup, ofstream& out){
+int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup){
try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(sharedfile); }
+ map<string, string> variables;
+ variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile));
+ variables["[extension]"] = m->getExtension(sharedfile);
+ variables["[tag]"] = lookup[0]->getLabel();
+ string outputFileName = getOutputFileName("shared", variables);
+ outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
vector<SharedRAbundVector> newRabunds; newRabunds.resize(lookup.size());
+ vector<string> headers;
for (int i = 0; i < lookup.size(); i++) {
newRabunds[i].setGroup(lookup[i]->getGroup());
newRabunds[i].setLabel(lookup[i]->getLabel());
for (int i = 0; i < lookup[0]->getNumBins(); i++) {
bool allZero = true;
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { out.close(); return 0; }
//for each group
for (int j = 0; j < lookup.size(); j++) {
//eliminates zero otus
if (allZero) { for (int j = 0; j < newRabunds.size(); j++) { newRabunds[j].pop_back(); } }
+ else { headers.push_back(m->currentBinLabels[i]); }
}
}else {
//for each otu
for (int i = 0; i < lookup[0]->getNumBins(); i++) {
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { out.close(); return 0; }
int totalAbund = 0;
//get total otu abundance
//eliminates otus below rare cutoff
if (totalAbund <= nseqs) { for (int j = 0; j < newRabunds.size(); j++) { newRabunds[j].pop_back(); } }
+ else { headers.push_back(m->currentBinLabels[i]); }
}
}
//do we have any otus above the rare cutoff
- if (newRabunds[0].getNumBins() != 0) {
+ if (newRabunds[0].getNumBins() != 0) {
+ out << "label\tGroup\tnumOtus\t";
+ for (int j = 0; j < headers.size(); j++) { out << headers[j] << '\t'; }
+ out << endl;
for (int j = 0; j < newRabunds.size(); j++) {
out << newRabunds[j].getLabel() << '\t' << newRabunds[j].getGroup() << '\t';
newRabunds[j].print(out);
}
}
+ out.close();
+
return 0;
}
catch(exception& e) {
projects / mothur.git / blobdiff