#include "subsamplecommand.h"
#include "sharedutilities.h"
#include "deconvolutecommand.h"
+#include "subsample.h"
//**********************************************************************************************************************
vector<string> SubSampleCommand::setParameters(){
exit(1);
}
}
+//**********************************************************************************************************************
+string SubSampleCommand::getOutputFileNameTag(string type, string inputName=""){
+ try {
+ string outputFileName = "";
+ map<string, vector<string> >::iterator it;
+
+ //is this a type this command creates
+ it = outputTypes.find(type);
+ if (it == outputTypes.end()) { m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+ else {
+ if (type == "fasta") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else if (type == "sabund") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else if (type == "name") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else if (type == "group") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else if (type == "list") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else if (type == "rabund") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else if (type == "shared") { outputFileName = "subsample" + m->getExtension(inputName); }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return outputFileName;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SubSampleCommand", "getOutputFileNameTag");
+ exit(1);
+ }
+}
+
//**********************************************************************************************************************
SubSampleCommand::SubSampleCommand(){
try {
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(fastafile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta", fastafile);
ofstream out;
m->openOutputFile(outputFileName, out);
if (namefile != "") {
m->mothurOut("Deconvoluting subsampled fasta file... "); m->mothurOutEndLine();
+ string outputNameFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + getOutputFileNameTag("name", namefile);
//use unique.seqs to create new name and fastafile
string inputString = "fasta=" + outputFileName;
m->mothurOut("/******************************************/"); m->mothurOutEndLine();
delete uniqueCommand;
m->mothurCalling = false;
- outputTypes["name"].push_back(filenames["name"][0]); outputNames.push_back(filenames["name"][0]);
- m->mothurRemove(outputFileName);
- outputFileName = filenames["fasta"][0];
-
+ m->renameFile(filenames["name"][0], outputNameFileName);
+ m->renameFile(filenames["fasta"][0], outputFileName);
+
+ outputTypes["name"].push_back(outputNameFileName); outputNames.push_back(outputNameFileName);
+
m->mothurOut("/******************************************/"); m->mothurOutEndLine();
m->mothurOut("Done."); m->mothurOutEndLine();
string groupOutputDir = outputDir;
if (outputDir == "") { groupOutputDir += m->hasPath(groupfile); }
- string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
+ string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
ofstream outGroup;
m->openOutputFile(groupOutputFileName, outGroup);
int SubSampleCommand::readNames() {
try {
- ifstream in;
- m->openInputFile(namefile, in);
-
- string thisname, repnames;
- map<string, vector<string> >::iterator it;
-
- while(!in.eof()){
-
- if (m->control_pressed) { in.close(); return 0; }
-
- in >> thisname; m->gobble(in); //read from first column
- in >> repnames; //read from second column
-
- it = nameMap.find(thisname);
- if (it == nameMap.end()) {
-
- vector<string> splitRepNames;
- m->splitAtComma(repnames, splitRepNames);
-
- nameMap[thisname] = splitRepNames;
- for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
-
- }else{ m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine(); }
-
- m->gobble(in);
- }
- in.close();
-
+ nameMap.clear();
+ m->readNames(namefile, nameMap);
+
+ //save names of all sequences
+ map<string, vector<string> >::iterator it;
+ for (it = nameMap.begin(); it != nameMap.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { names.push_back((it->second)[i]); } }
+
return 0;
}
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(sharedfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + ".subsample" + m->getExtension(sharedfile);
-
-
- ofstream out;
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + "." +getOutputFileNameTag("shared", sharedfile);
+ SubSample sample;
+ vector<string> subsampledLabels = sample.getSample(thislookup, size);
+
+ if (m->control_pressed) { return 0; }
+
+ ofstream out;
m->openOutputFile(outputFileName, out);
outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
- int numBins = thislookup[0]->getNumBins();
- for (int i = 0; i < thislookup.size(); i++) {
- int thisSize = thislookup[i]->getNumSeqs();
-
- if (thisSize != size) {
-
- string thisgroup = thislookup[i]->getGroup();
-
- OrderVector* order = new OrderVector();
- for(int p=0;p<numBins;p++){
- for(int j=0;j<thislookup[i]->getAbundance(p);j++){
- order->push_back(p);
- }
- }
- random_shuffle(order->begin(), order->end());
-
- SharedRAbundVector* temp = new SharedRAbundVector(numBins);
- temp->setLabel(thislookup[i]->getLabel());
- temp->setGroup(thislookup[i]->getGroup());
-
- delete thislookup[i];
- thislookup[i] = temp;
-
-
- for (int j = 0; j < size; j++) {
-
- if (m->control_pressed) { delete order; out.close(); return 0; }
-
- //get random number to sample from order between 0 and thisSize-1.
- //don't need this because of the random shuffle above
- //int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
-
- int bin = order->get(j);
-
- int abund = thislookup[i]->getAbundance(bin);
- thislookup[i]->set(bin, (abund+1), thisgroup);
- }
- delete order;
- }
- }
-
- //subsampling may have created some otus with no sequences in them
- eliminateZeroOTUS(thislookup);
-
- if (m->control_pressed) { out.close(); return 0; }
-
+ m->currentBinLabels = subsampledLabels;
+
thislookup[0]->printHeaders(out);
for (int i = 0; i < thislookup.size(); i++) {
out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
thislookup[i]->print(out);
}
-
out.close();
-
- //save mothurOut's binLabels to restore for next label
+
+
+ //save mothurOut's binLabels to restore for next label
m->currentBinLabels = saveBinLabels;
return 0;
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + getOutputFileNameTag("list", listfile);
ofstream out;
m->openOutputFile(outputFileName, out);
outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(rabundfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + getOutputFileNameTag("rabund", rabundfile);
ofstream out;
m->openOutputFile(outputFileName, out);
outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
string thisOutputDir = outputDir;
if (outputDir == "") { thisOutputDir += m->hasPath(sabundfile); }
- string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
-
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + getOutputFileNameTag("sabund", sabundfile);
ofstream out;
m->openOutputFile(outputFileName, out);
outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
}
}
//**********************************************************************************************************************
-int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
- try {
-
- vector<SharedRAbundVector*> newLookup;
- for (int i = 0; i < thislookup.size(); i++) {
- SharedRAbundVector* temp = new SharedRAbundVector();
- temp->setLabel(thislookup[i]->getLabel());
- temp->setGroup(thislookup[i]->getGroup());
- newLookup.push_back(temp);
- }
-
- //for each bin
- vector<string> newBinLabels;
- string snumBins = toString(thislookup[0]->getNumBins());
- for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
- if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) { delete newLookup[j]; } return 0; }
-
- //look at each sharedRabund and make sure they are not all zero
- bool allZero = true;
- for (int j = 0; j < thislookup.size(); j++) {
- if (thislookup[j]->getAbundance(i) != 0) { allZero = false; break; }
- }
-
- //if they are not all zero add this bin
- if (!allZero) {
- for (int j = 0; j < thislookup.size(); j++) {
- newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
- }
- //if there is a bin label use it otherwise make one
- string binLabel = "Otu";
- string sbinNumber = toString(i+1);
- if (sbinNumber.length() < snumBins.length()) {
- int diff = snumBins.length() - sbinNumber.length();
- for (int h = 0; h < diff; h++) { binLabel += "0"; }
- }
- binLabel += sbinNumber;
- if (i < m->currentBinLabels.size()) { binLabel = m->currentBinLabels[i]; }
-
- newBinLabels.push_back(binLabel);
- }
- }
-
- for (int j = 0; j < thislookup.size(); j++) { delete thislookup[j]; }
- thislookup.clear();
-
- thislookup = newLookup;
- m->currentBinLabels = newBinLabels;
-
- return 0;
-
- }
- catch(exception& e) {
- m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
projects / mothur.git / blobdiff