#include "sharedsobscollectsummary.h"
#include "sharedchao1.h"
#include "sharedace.h"
+#include "sharednseqs.h"
#include "sharedjabund.h"
#include "sharedsorabund.h"
#include "sharedjclass.h"
#include "sharedsorest.h"
#include "sharedthetayc.h"
#include "sharedthetan.h"
+#include "sharedkstest.h"
+#include "whittaker.h"
+#include "sharedochiai.h"
+#include "sharedanderbergs.h"
+#include "sharedkulczynski.h"
+#include "sharedkulczynskicody.h"
+#include "sharedlennon.h"
+#include "sharedmorisitahorn.h"
+#include "sharedbraycurtis.h"
+#include "sharedjackknife.h"
+#include "whittaker.h"
//**********************************************************************************************************************
openOutputFile(outputFileName, outputFileHandle);
format = globaldata->getFormat();
validCalculator = new ValidCalculators();
+ mult = false;
int i;
for (i=0; i<globaldata->Estimators.size(); i++) {
sumCalculators.push_back(new SharedChao1());
}else if (globaldata->Estimators[i] == "sharedace") {
sumCalculators.push_back(new SharedAce());
- }else if (globaldata->Estimators[i] == "sharedjabund") {
- sumCalculators.push_back(new SharedJAbund());
- }else if (globaldata->Estimators[i] == "sharedsorensonabund") {
- sumCalculators.push_back(new SharedSorAbund());
- }else if (globaldata->Estimators[i] == "sharedjclass") {
- sumCalculators.push_back(new SharedJclass());
- }else if (globaldata->Estimators[i] == "sharedsorclass") {
- sumCalculators.push_back(new SharedSorClass());
- }else if (globaldata->Estimators[i] == "sharedjest") {
- sumCalculators.push_back(new SharedJest());
- }else if (globaldata->Estimators[i] == "sharedsorest") {
- sumCalculators.push_back(new SharedSorEst());
- }else if (globaldata->Estimators[i] == "sharedthetayc") {
- sumCalculators.push_back(new SharedThetaYC());
- }else if (globaldata->Estimators[i] == "sharedthetan") {
- sumCalculators.push_back(new SharedThetaN());
+ }else if (globaldata->Estimators[i] == "jabund") {
+ sumCalculators.push_back(new JAbund());
+ }else if (globaldata->Estimators[i] == "sorabund") {
+ sumCalculators.push_back(new SorAbund());
+ }else if (globaldata->Estimators[i] == "jclass") {
+ sumCalculators.push_back(new Jclass());
+ }else if (globaldata->Estimators[i] == "sorclass") {
+ sumCalculators.push_back(new SorClass());
+ }else if (globaldata->Estimators[i] == "jest") {
+ sumCalculators.push_back(new Jest());
+ }else if (globaldata->Estimators[i] == "sorest") {
+ sumCalculators.push_back(new SorEst());
+ }else if (globaldata->Estimators[i] == "thetayc") {
+ sumCalculators.push_back(new ThetaYC());
+ }else if (globaldata->Estimators[i] == "thetan") {
+ sumCalculators.push_back(new ThetaN());
+ }else if (globaldata->Estimators[i] == "kstest") {
+ sumCalculators.push_back(new KSTest());
+ }else if (globaldata->Estimators[i] == "sharednseqs") {
+ sumCalculators.push_back(new SharedNSeqs());
+ }else if (globaldata->Estimators[i] == "ochiai") {
+ sumCalculators.push_back(new Ochiai());
+ }else if (globaldata->Estimators[i] == "anderberg") {
+ sumCalculators.push_back(new Anderberg());
+ }else if (globaldata->Estimators[i] == "kulczynski") {
+ sumCalculators.push_back(new Kulczynski());
+ }else if (globaldata->Estimators[i] == "kulczynskicody") {
+ sumCalculators.push_back(new KulczynskiCody());
+ }else if (globaldata->Estimators[i] == "lennon") {
+ sumCalculators.push_back(new Lennon());
+ }else if (globaldata->Estimators[i] == "morisitahorn") {
+ sumCalculators.push_back(new MorHorn());
+ }else if (globaldata->Estimators[i] == "braycurtis") {
+ sumCalculators.push_back(new BrayCurtis());
+ }else if (globaldata->Estimators[i] == "whittaker") {
+ sumCalculators.push_back(new Whittaker());
}
}
}
-
//reset calc for next command
globaldata->setCalc("");
int SummarySharedCommand::execute(){
try {
int count = 1;
-
+
//if the users entered no valid calculators don't execute command
if (sumCalculators.size() == 0) { return 0; }
-
- if (format == "sharedfile") {
- read = new ReadPhilFile(globaldata->inputFileName);
- read->read(&*globaldata);
-
- input = globaldata->ginput;
- order = input->getSharedOrderVector();
- }else {
- //you are using a list and a groupfile
- read = new ReadPhilFile(globaldata->inputFileName);
- read->read(&*globaldata);
-
- input = globaldata->ginput;
- SharedList = globaldata->gSharedList;
- order = SharedList->getSharedOrderVector();
+ //check if any calcs can do multiples
+ else{
+ for (int i = 0; i < sumCalculators.size(); i++) {
+ if (sumCalculators[i]->getMultiple() == true) { mult = true; }
+ }
}
+ //read first line
+ read = new ReadOTUFile(globaldata->inputFileName);
+ read->read(&*globaldata);
+
+ input = globaldata->ginput;
+ lookup = input->getSharedRAbundVectors();
+ vector<SharedRAbundVector*> lastLookup = lookup;
+
//output estimator names as column headers
outputFileHandle << "label" <<'\t' << "comparison" << '\t';
for(int i=0;i<sumCalculators.size();i++){
}
outputFileHandle << endl;
- //set users groups
- setGroups();
-
- while(order != NULL){
+ //create file and put column headers for multiple groups file
+ if (mult == true) {
+ outAllFileName = ((getRootName(globaldata->inputFileName)) + "sharedmultiple.summary");
+ openOutputFile(outAllFileName, outAll);
+
+ outAll << "label" <<'\t' << "comparison" << '\t';
+ for(int i=0;i<sumCalculators.size();i++){
+ if (sumCalculators[i]->getMultiple() == true) {
+ outAll << '\t' << sumCalculators[i]->getName();
+ }
+ }
+ outAll << endl;
+ }
- if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
-
- cout << order->getLabel() << '\t' << count << endl;
- getSharedVectors(); //fills group vectors from order vector.
+ if (lookup.size() < 2) {
+ cout << "I cannot run the command without at least 2 valid groups.";
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+
+ //close files and clean up
+ outputFileHandle.close(); remove(outputFileName.c_str());
+ if (mult == true) { outAll.close(); remove(outAllFileName.c_str()); }
+ return 0;
+ }
+
+ //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;
+
+ //as long as you are not at the end of the file or done wih the lines you want
+ while((lookup[0] != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
+
+ if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(lookup[0]->getLabel()) == 1){
+ cout << lookup[0]->getLabel() << '\t' << count << endl;
+ process(lookup);
- //randomize group order
- if (globaldata->getJumble() == "1") { random_shuffle(lookup.begin(), lookup.end()); }
-
- int n = 1;
- for (int k = 0; k < (lookup.size() - 1); k++) { // pass cdd each set of groups to commpare
- for (int l = n; l < lookup.size(); l++) {
- outputFileHandle << order->getLabel() << '\t' << (lookup[k]->getGroup() + lookup[l]->getGroup()) << '\t' << '\t'; //print out label and group
- for(int i=0;i<sumCalculators.size();i++){
- sumCalculators[i]->getValues(lookup[k], lookup[l]); //saves the calculator outputs
- outputFileHandle << '\t';
- sumCalculators[i]->print(outputFileHandle);
- }
- outputFileHandle << endl;
- }
- n++;
- }
+ processedLabels.insert(lookup[0]->getLabel());
+ userLabels.erase(lookup[0]->getLabel());
+ userLines.erase(count);
}
+
+ if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLookup[0]->getLabel()) != 1)) {
+ cout << lastLookup[0]->getLabel() << '\t' << count << endl;
+ process(lastLookup);
+
+ processedLabels.insert(lastLookup[0]->getLabel());
+ userLabels.erase(lastLookup[0]->getLabel());
+ }
+
+ //prevent memory leak
+ if (count != 1) { for (int i = 0; i < lastLookup.size(); i++) { delete lastLookup[i]; } }
+ lastLookup = lookup;
+
//get next line to process
- if (format == "sharedfile") {
- order = input->getSharedOrderVector();
+ lookup = input->getSharedRAbundVectors();
+ 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(lastLookup[0]->getLabel()) != 1) {
+ cout << ". I will use " << lastLookup[0]->getLabel() << "." << endl;
+ needToRun = true;
}else {
- //you are using a list and a groupfile
- SharedList = input->getSharedListVector(); //get new list vector to process
- if (SharedList != NULL) {
- order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
- }else {
- break;
- }
+ cout << ". Please refer to " << lastLookup[0]->getLabel() << "." << endl;
}
- count++;
}
+ //run last line if you need to
+ if (needToRun == true) {
+ cout << lastLookup[0]->getLabel() << '\t' << count << endl;
+ process(lastLookup);
+ }
+
+ for (int i = 0; i < lastLookup.size(); i++) { delete lastLookup[i]; }
+
//reset groups parameter
globaldata->Groups.clear(); globaldata->setGroups("");
+
+ //close files
+ outputFileHandle.close();
+ if (mult == true) { outAll.close(); }
return 0;
}
}
}
-//**********************************************************************************************************************
-
-void SummarySharedCommand::getSharedVectors(){
-try {
- lookup.clear();
- //create and initialize vector of sharedvectors, one for each group
- for (int i = 0; i < globaldata->gGroupmap->getNumGroups(); i++) {
- SharedRAbundVector* temp = new SharedRAbundVector(order->getNumBins());
- temp->setLabel(order->getLabel());
- temp->setGroup(globaldata->gGroupmap->namesOfGroups[i]);
- lookup.push_back(temp);
- }
-
- int numSeqs = order->size();
- //sample all the members
- for(int i=0;i<numSeqs;i++){
- //get first sample
- individual chosen = order->get(i);
- int abundance;
-
- //set info for sharedvector in chosens group
- for (int j = 0; j < lookup.size(); j++) {
- if (chosen.group == lookup[j]->getGroup()) {
- abundance = lookup[j]->getAbundance(chosen.bin);
- lookup[j]->set(chosen.bin, (abundance + 1), chosen.group);
- break;
- }
- }
- }
-
- //get rid of vectors from groups you don't want to analyze
- for (int r = 0; r < lookup.size(); r++) {
- if (inUsersGroups(lookup[r]->getGroup(), globaldata->Groups) != true) {
- lookup.erase(lookup.begin()+r);
- }
- }
-
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the SummarySharedCommand class function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-
-}
-
-//**********************************************************************************************************************
-void SummarySharedCommand::setGroups() {
+/***********************************************************/
+void SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup) {
try {
- //if the user has not entered specific groups to analyze then do them all
- if (globaldata->Groups.size() != 0) {
- if (globaldata->Groups[0] != "all") {
- //check that groups are valid
- for (int i = 0; i < globaldata->Groups.size(); i++) {
- if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
- cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
- // erase the invalid group from globaldata->Groups
- globaldata->Groups.erase(globaldata->Groups.begin()+i);
+ //loop through calculators and add to file all for all calcs that can do mutiple groups
+ if (mult == true) {
+ //output label
+ outAll << thisLookup[0]->getLabel() << '\t';
+
+ //output groups names
+ string outNames = "";
+ for (int j = 0; j < thisLookup.size(); j++) {
+ outNames += thisLookup[j]->getGroup() + "-";
}
- }
-
- //if the user only entered invalid groups
- if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
- cout << "When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile." << endl;
- for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
- globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
+ outNames = outNames.substr(0, outNames.length()-1); //rip off extra '-';
+ outAll << outNames << '\t';
+
+ for(int i=0;i<sumCalculators.size();i++){
+ if (sumCalculators[i]->getMultiple() == true) {
+ sumCalculators[i]->getValues(thisLookup);
+ outAll << '\t';
+ sumCalculators[i]->print(outAll);
+ }
}
+ outAll << endl;
}
- }else{//user has enter "all" and wants the default groups
- globaldata->Groups.clear();
- for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
- globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
+
+ int n = 1;
+ vector<SharedRAbundVector*> subset;
+ for (int k = 0; k < (thisLookup.size() - 1); k++) { // pass cdd each set of groups to commpare
+ for (int l = n; l < thisLookup.size(); l++) {
+
+ outputFileHandle << thisLookup[0]->getLabel() << '\t';
+
+ subset.clear(); //clear out old pair of sharedrabunds
+ //add new pair of sharedrabunds
+ subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
+
+ //sort groups to be alphanumeric
+ if (thisLookup[k]->getGroup() > thisLookup[l]->getGroup()) {
+ outputFileHandle << (thisLookup[l]->getGroup() +'\t' + thisLookup[k]->getGroup()) << '\t'; //print out groups
+ }else{
+ outputFileHandle << (thisLookup[k]->getGroup() +'\t' + thisLookup[l]->getGroup()) << '\t'; //print out groups
+ }
+
+ for(int i=0;i<sumCalculators.size();i++) {
+
+ sumCalculators[i]->getValues(subset); //saves the calculator outputs
+ outputFileHandle << '\t';
+ sumCalculators[i]->print(outputFileHandle);
+ }
+ outputFileHandle << endl;
+ }
+ n++;
}
- globaldata->setGroups("");
- }
- }else {
- for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
- globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
- }
- }
-
+
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
exit(1);
}
catch(...) {
- cout << "An unknown error has occurred in the SummarySharedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ cout << "An unknown error has occurred in the SummarySharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
exit(1);
}
-
}
-/***********************************************************/
+
+/***********************************************************/
\ No newline at end of file
projects / mothur.git / blobdiff