5 * Created by Sarah Westcott on 3/30/09.
6 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
10 #include "venncommand.h"
14 //#include "jackknife.h"
15 #include "sharedsobscollectsummary.h"
16 #include "sharedchao1.h"
17 #include "sharedace.h"
21 //**********************************************************************************************************************
22 vector<string> VennCommand::getValidParameters(){
24 string Array[] = {"groups","label","calc","permute", "abund","nseqs","outputdir","inputdir"};
25 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
29 m->errorOut(e, "VennCommand", "getValidParameters");
33 //**********************************************************************************************************************
34 VennCommand::VennCommand(){
37 //initialize outputTypes
38 vector<string> tempOutNames;
39 outputTypes["svg"] = tempOutNames;
42 m->errorOut(e, "VennCommand", "VennCommand");
46 //**********************************************************************************************************************
47 vector<string> VennCommand::getRequiredParameters(){
49 vector<string> myArray;
53 m->errorOut(e, "VennCommand", "getRequiredParameters");
57 //**********************************************************************************************************************
58 vector<string> VennCommand::getRequiredFiles(){
60 string Array[] = {"list","shared","or"};
61 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
65 m->errorOut(e, "VennCommand", "getRequiredFiles");
69 //**********************************************************************************************************************
71 VennCommand::VennCommand(string option) {
73 globaldata = GlobalData::getInstance();
78 //allow user to run help
79 if(option == "help") { help(); abort = true; }
82 //valid paramters for this command
83 string AlignArray[] = {"groups","label","calc","permute", "abund","nseqs","outputdir","inputdir"};
84 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
86 OptionParser parser(option);
87 map<string,string> parameters = parser.getParameters();
89 ValidParameters validParameter;
91 //check to make sure all parameters are valid for command
92 for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
93 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
96 //make sure the user has already run the read.otu command
97 if ((globaldata->getListFile() == "") && (globaldata->getSharedFile() == "")) {
98 m->mothurOut("You must read a list, or a list and a group, or a shared before you can use the venn command."); m->mothurOutEndLine(); abort = true;
101 //if the user changes the output directory command factory will send this info to us in the output parameter
102 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
104 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
107 //check for optional parameter and set defaults
108 // ...at some point should added some additional type checking...
109 label = validParameter.validFile(parameters, "label", false);
110 if (label == "not found") { label = ""; }
112 if(label != "all") { m->splitAtDash(label, labels); allLines = 0; }
113 else { allLines = 1; }
116 //if the user has not specified any labels use the ones from read.otu
118 allLines = globaldata->allLines;
119 labels = globaldata->labels;
122 groups = validParameter.validFile(parameters, "groups", false);
123 if (groups == "not found") { groups = ""; }
125 m->splitAtDash(groups, Groups);
126 globaldata->Groups = Groups;
129 format = globaldata->getFormat();
130 calc = validParameter.validFile(parameters, "calc", false);
131 if (calc == "not found") {
132 if(format == "list") { calc = "sobs"; }
133 else { calc = "sharedsobs"; }
136 if (calc == "default") {
137 if(format == "list") { calc = "sobs"; }
138 else { calc = "sharedsobs"; }
141 m->splitAtDash(calc, Estimators);
144 temp = validParameter.validFile(parameters, "abund", false); if (temp == "not found") { temp = "10"; }
145 convert(temp, abund);
147 temp = validParameter.validFile(parameters, "nseqs", false); if (temp == "not found"){ temp = "f"; }
148 nseqs = m->isTrue(temp);
150 temp = validParameter.validFile(parameters, "permute", false); if (temp == "not found"){ temp = "f"; }
151 perm = m->isTrue(temp);
153 if (abort == false) {
154 validCalculator = new ValidCalculators();
158 if (format == "list") {
159 for (i=0; i<Estimators.size(); i++) {
160 if (validCalculator->isValidCalculator("vennsingle", Estimators[i]) == true) {
161 if (Estimators[i] == "sobs") {
162 vennCalculators.push_back(new Sobs());
163 }else if (Estimators[i] == "chao") {
164 vennCalculators.push_back(new Chao1());
165 }else if (Estimators[i] == "ace") {
168 vennCalculators.push_back(new Ace(abund));
173 for (i=0; i<Estimators.size(); i++) {
174 if (validCalculator->isValidCalculator("vennshared", Estimators[i]) == true) {
175 if (Estimators[i] == "sharedsobs") {
176 vennCalculators.push_back(new SharedSobsCS());
177 }else if (Estimators[i] == "sharedchao") {
178 vennCalculators.push_back(new SharedChao1());
179 }else if (Estimators[i] == "sharedace") {
180 vennCalculators.push_back(new SharedAce());
186 //if the users entered no valid calculators don't execute command
187 if (vennCalculators.size() == 0) { m->mothurOut("No valid calculators given, please correct."); m->mothurOutEndLine(); abort = true; }
188 else { venn = new Venn(outputDir, nseqs); }
196 catch(exception& e) {
197 m->errorOut(e, "VennCommand", "VennCommand");
202 //**********************************************************************************************************************
204 void VennCommand::help(){
206 m->mothurOut("The venn command can only be executed after a successful read.otu command.\n");
207 m->mothurOut("The venn command parameters are groups, calc, abund, nseqs, permute and label. No parameters are required.\n");
208 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included in your venn diagram, you may only use a maximum of 4 groups.\n");
209 m->mothurOut("The group names are separated by dashes. The label allows you to select what distance levels you would like a venn diagram created for, and are also separated by dashes.\n");
210 m->mothurOut("The venn command should be in the following format: venn(groups=yourGroups, calc=yourCalcs, label=yourLabels, abund=yourAbund).\n");
211 m->mothurOut("Example venn(groups=A-B-C, calc=sharedsobs-sharedchao, abund=20).\n");
212 m->mothurOut("The default value for groups is all the groups in your groupfile up to 4, and all labels in your inputfile will be used.\n");
213 m->mothurOut("The default value for calc is sobs if you have only read a list file or if you have selected only one group, and sharedsobs if you have multiple groups.\n");
214 m->mothurOut("The default available estimators for calc are sobs, chao and ace if you have only read a list file, and sharedsobs, sharedchao and sharedace if you have read a list and group file or a shared file.\n");
215 m->mothurOut("The nseqs parameter will output the number of sequences represented by the otus in the picture, default=F.\n");
216 m->mothurOut("If you have more than 4 groups, the permute parameter will find all possible combos of 4 of your groups and create pictures for them, default=F.\n");
217 m->mothurOut("The only estimators available four 4 groups are sharedsobs and sharedchao.\n");
218 m->mothurOut("The venn command outputs a .svg file for each calculator you specify at each distance you choose.\n");
219 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
221 catch(exception& e) {
222 m->errorOut(e, "VennCommand", "help");
228 //**********************************************************************************************************************
230 VennCommand::~VennCommand(){
231 if (abort == false) {
232 delete input; globaldata->ginput = NULL;
235 globaldata->sabund = NULL;
236 delete validCalculator;
241 //**********************************************************************************************************************
243 int VennCommand::execute(){
246 if (abort == true) { return 0; }
250 if (format == "sharedfile") {
252 read = new ReadOTUFile(globaldata->inputFileName);
253 read->read(&*globaldata);
255 input = globaldata->ginput;
256 lookup = input->getSharedRAbundVectors();
257 lastLabel = lookup[0]->getLabel();
259 if ((lookup.size() > 4) && (perm)) { combosOfFour = findCombinations(lookup.size()); }
260 }else if (format == "list") {
261 //you are using just a list file and have only one group
262 read = new ReadOTUFile(globaldata->inputFileName);
263 read->read(&*globaldata);
265 sabund = globaldata->sabund;
266 lastLabel = sabund->getLabel();
267 input = globaldata->ginput;
270 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
271 set<string> processedLabels;
272 set<string> userLabels = labels;
274 if (format != "list") {
276 //as long as you are not at the end of the file or done wih the lines you want
277 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
279 if (m->control_pressed) {
280 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
281 for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
282 globaldata->Groups.clear();
283 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
287 if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
288 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
289 processedLabels.insert(lookup[0]->getLabel());
290 userLabels.erase(lookup[0]->getLabel());
292 if ((lookup.size() > 4) && (!perm)){
293 m->mothurOut("Error: Too many groups chosen. You may use up to 4 groups with the venn command. I will use the first four groups in your groupfile. If you set perm=t, I will find all possible combos of 4 groups."); m->mothurOutEndLine();
294 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
296 vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
297 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
299 }else if ((lookup.size() > 4) && (perm)) {
300 set< set<int> >::iterator it3;
301 set<int>::iterator it2;
302 for (it3 = combosOfFour.begin(); it3 != combosOfFour.end(); it3++) {
304 set<int> poss = *it3;
305 vector<SharedRAbundVector*> subset;
306 for (it2 = poss.begin(); it2 != poss.end(); it2++) { subset.push_back(lookup[*it2]); }
308 vector<string> outfilenames = venn->getPic(subset, vennCalculators);
309 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
312 vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
313 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
317 if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
318 string saveLabel = lookup[0]->getLabel();
320 for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
321 lookup = input->getSharedRAbundVectors(lastLabel);
323 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
324 processedLabels.insert(lookup[0]->getLabel());
325 userLabels.erase(lookup[0]->getLabel());
327 if ((lookup.size() > 4) && (!perm)){
328 m->mothurOut("Error: Too many groups chosen. You may use up to 4 groups with the venn command. I will use the first four groups in your groupfile. If you set perm=t, I will find all possible combos of 4 groups."); m->mothurOutEndLine();
329 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
331 vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
332 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
334 }else if ((lookup.size() > 4) && (perm)) {
335 set< set<int> >::iterator it3;
336 set<int>::iterator it2;
337 for (it3 = combosOfFour.begin(); it3 != combosOfFour.end(); it3++) {
339 set<int> poss = *it3;
340 vector<SharedRAbundVector*> subset;
341 for (it2 = poss.begin(); it2 != poss.end(); it2++) { subset.push_back(lookup[*it2]); }
343 vector<string> outfilenames = venn->getPic(subset, vennCalculators);
344 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
347 vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
348 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
351 //restore real lastlabel to save below
352 lookup[0]->setLabel(saveLabel);
356 lastLabel = lookup[0]->getLabel();
358 //get next line to process
359 for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
360 lookup = input->getSharedRAbundVectors();
363 if (m->control_pressed) {
364 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
365 globaldata->Groups.clear();
366 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
371 //output error messages about any remaining user labels
372 set<string>::iterator it;
373 bool needToRun = false;
374 for (it = userLabels.begin(); it != userLabels.end(); it++) {
375 m->mothurOut("Your file does not include the label " + *it);
376 if (processedLabels.count(lastLabel) != 1) {
377 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
380 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
384 //run last label if you need to
385 if (needToRun == true) {
386 for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
387 lookup = input->getSharedRAbundVectors(lastLabel);
389 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
390 processedLabels.insert(lookup[0]->getLabel());
391 userLabels.erase(lookup[0]->getLabel());
393 if ((lookup.size() > 4) && (!perm)){
394 m->mothurOut("Error: Too many groups chosen. You may use up to 4 groups with the venn command. I will use the first four groups in your groupfile. If you set perm=t, I will find all possible combos of 4 groups."); m->mothurOutEndLine();
395 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
397 vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
398 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
400 }else if ((lookup.size() > 4) && (perm)) {
401 set< set<int> >::iterator it3;
402 set<int>::iterator it2;
403 for (it3 = combosOfFour.begin(); it3 != combosOfFour.end(); it3++) {
405 set<int> poss = *it3;
406 vector<SharedRAbundVector*> subset;
407 for (it2 = poss.begin(); it2 != poss.end(); it2++) { subset.push_back(lookup[*it2]); }
409 vector<string> outfilenames = venn->getPic(subset, vennCalculators);
410 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
413 vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
414 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
417 for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
421 //reset groups parameter
422 globaldata->Groups.clear();
424 if (m->control_pressed) {
425 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
426 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
433 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
435 if (m->control_pressed) {
436 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
438 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
442 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
444 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
445 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
446 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
449 processedLabels.insert(sabund->getLabel());
450 userLabels.erase(sabund->getLabel());
453 if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
454 string saveLabel = sabund->getLabel();
457 sabund = input->getSAbundVector(lastLabel);
459 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
460 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
461 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
464 processedLabels.insert(sabund->getLabel());
465 userLabels.erase(sabund->getLabel());
467 //restore real lastlabel to save below
468 sabund->setLabel(saveLabel);
471 lastLabel = sabund->getLabel();
474 sabund = input->getSAbundVector();
477 if (m->control_pressed) {
478 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
479 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
483 //output error messages about any remaining user labels
484 set<string>::iterator it;
485 bool needToRun = false;
486 for (it = userLabels.begin(); it != userLabels.end(); it++) {
487 m->mothurOut("Your file does not include the label " + *it);
488 if (processedLabels.count(lastLabel) != 1) {
489 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
492 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
496 //run last label if you need to
497 if (needToRun == true) {
498 if (sabund != NULL) { delete sabund; }
499 sabund = input->getSAbundVector(lastLabel);
501 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
502 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
503 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
509 if (m->control_pressed) {
510 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
511 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
516 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
518 m->mothurOutEndLine();
519 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
520 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
521 m->mothurOutEndLine();
526 catch(exception& e) {
527 m->errorOut(e, "VennCommand", "execute");
531 //**********************************************************************************************************************
532 //returns a vector of sets containing the 4 group combinations
533 set< set<int> > VennCommand::findCombinations(int lookupSize){
535 set< set<int> > combos;
538 for (int i = 0; i < lookupSize; i++) { possibles.insert(i); }
540 getCombos(possibles, combos);
545 catch(exception& e) {
546 m->errorOut(e, "VennCommand", "findCombinations");
550 //**********************************************************************************************************************
551 //recusively finds combos of 4
552 int VennCommand::getCombos(set<int> possibles, set< set<int> >& combos){
555 if (possibles.size() == 4) { //done
556 if (combos.count(possibles) == 0) { //no dups
557 combos.insert(possibles);
559 }else { //we still have work to do
560 set<int>::iterator it;
561 set<int>::iterator it2;
562 for (it = possibles.begin(); it != possibles.end(); it++) {
564 set<int> newPossibles;
565 for (it2 = possibles.begin(); it2 != possibles.end(); it2++) { //all possible combos of one length smaller
567 newPossibles.insert(*it2);
570 getCombos(newPossibles, combos);
576 catch(exception& e) {
577 m->errorOut(e, "VennCommand", "getCombos");
582 //**********************************************************************************************************************