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"
20 //**********************************************************************************************************************
21 vector<string> VennCommand::setParameters(){
23 CommandParameter plist("list", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(plist);
24 CommandParameter pshared("shared", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(pshared);
25 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
26 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
27 CommandParameter pcalc("calc", "String", "", "", "", "", "",false,false); parameters.push_back(pcalc);
28 CommandParameter pabund("abund", "Number", "", "10", "", "", "",false,false); parameters.push_back(pabund);
29 CommandParameter pnseqs("nseqs", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pnseqs);
30 CommandParameter pfontsize("fontsize", "Number", "", "24", "", "", "",false,false); parameters.push_back(pfontsize);
31 CommandParameter ppermute("permute", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppermute);
32 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
33 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
35 vector<string> myArray;
36 for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
40 m->errorOut(e, "VennCommand", "setParameters");
44 //**********************************************************************************************************************
45 string VennCommand::getHelpString(){
47 string helpString = "";
48 helpString += "The venn command parameters are list, shared, groups, calc, abund, nseqs, permute, fontsize and label. shared, relabund, list, rabund or sabund is required unless you have a valid current file.\n";
49 helpString += "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";
50 helpString += "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";
51 helpString += "The fontsize parameter allows you to adjust the font size of the picture created, default=24.\n";
52 helpString += "The venn command should be in the following format: venn(groups=yourGroups, calc=yourCalcs, label=yourLabels, abund=yourAbund).\n";
53 helpString += "Example venn(groups=A-B-C, calc=sharedsobs-sharedchao, abund=20).\n";
54 helpString += "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";
55 helpString += "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";
56 helpString += "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 shared file.\n";
57 helpString += "The nseqs parameter will output the number of sequences represented by the otus in the picture, default=F.\n";
58 helpString += "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";
59 helpString += "The only estimators available four 4 groups are sharedsobs and sharedchao.\n";
60 helpString += "The venn command outputs a .svg file for each calculator you specify at each distance you choose.\n";
61 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
65 m->errorOut(e, "VennCommand", "getHelpString");
70 //**********************************************************************************************************************
71 VennCommand::VennCommand(){
73 abort = true; calledHelp = true;
75 vector<string> tempOutNames;
76 outputTypes["svg"] = tempOutNames;
79 m->errorOut(e, "VennCommand", "VennCommand");
83 //**********************************************************************************************************************
85 VennCommand::VennCommand(string option) {
87 abort = false; calledHelp = false;
90 //allow user to run help
91 if(option == "help") { help(); abort = true; calledHelp = true; }
92 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
95 vector<string> myArray = setParameters();
97 OptionParser parser(option);
98 map<string,string> parameters = parser.getParameters();
99 map<string,string>::iterator it;
101 ValidParameters validParameter;
103 //check to make sure all parameters are valid for command
104 for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
105 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
108 //if the user changes the input directory command factory will send this info to us in the output parameter
109 string inputDir = validParameter.validFile(parameters, "inputdir", false);
110 if (inputDir == "not found"){ inputDir = ""; }
113 it = parameters.find("shared");
114 //user has given a template file
115 if(it != parameters.end()){
116 path = m->hasPath(it->second);
117 //if the user has not given a path then, add inputdir. else leave path alone.
118 if (path == "") { parameters["shared"] = inputDir + it->second; }
121 it = parameters.find("list");
122 //user has given a template file
123 if(it != parameters.end()){
124 path = m->hasPath(it->second);
125 //if the user has not given a path then, add inputdir. else leave path alone.
126 if (path == "") { parameters["list"] = inputDir + it->second; }
130 //check for required parameters
131 listfile = validParameter.validFile(parameters, "list", true);
132 if (listfile == "not open") { listfile = ""; abort = true; }
133 else if (listfile == "not found") { listfile = ""; }
134 else { format = "list"; inputfile = listfile; m->setListFile(listfile); }
136 sharedfile = validParameter.validFile(parameters, "shared", true);
137 if (sharedfile == "not open") { sharedfile = ""; abort = true; }
138 else if (sharedfile == "not found") { sharedfile = ""; }
139 else { format = "sharedfile"; inputfile = sharedfile; m->setSharedFile(sharedfile); }
141 if ((sharedfile == "") && (listfile == "")) {
142 //is there are current file available for any of these?
143 //give priority to shared, then list, then rabund, then sabund
144 //if there is a current shared file, use it
145 sharedfile = m->getSharedFile();
146 if (sharedfile != "") { inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
148 listfile = m->getListFile();
149 if (listfile != "") { inputfile = listfile; format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
151 m->mothurOut("No valid current files. You must provide a list or shared file."); m->mothurOutEndLine();
157 //if the user changes the output directory command factory will send this info to us in the output parameter
158 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = m->hasPath(inputfile); }
160 //check for optional parameter and set defaults
161 // ...at some point should added some additional type checking...
162 label = validParameter.validFile(parameters, "label", false);
163 if (label == "not found") { label = ""; }
165 if(label != "all") { m->splitAtDash(label, labels); allLines = 0; }
166 else { allLines = 1; }
169 groups = validParameter.validFile(parameters, "groups", false);
170 if (groups == "not found") { groups = ""; }
172 m->splitAtDash(groups, Groups);
173 m->setGroups(Groups);
176 calc = validParameter.validFile(parameters, "calc", false);
177 if (calc == "not found") {
178 if(format == "list") { calc = "sobs"; }
179 else { calc = "sharedsobs"; }
182 if (calc == "default") {
183 if(format == "list") { calc = "sobs"; }
184 else { calc = "sharedsobs"; }
187 m->splitAtDash(calc, Estimators);
188 if (m->inUsersGroups("citation", Estimators)) {
189 ValidCalculators validCalc; validCalc.printCitations(Estimators);
190 //remove citation from list of calcs
191 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") { Estimators.erase(Estimators.begin()+i); break; } }
195 temp = validParameter.validFile(parameters, "abund", false); if (temp == "not found") { temp = "10"; }
196 m->mothurConvert(temp, abund);
198 temp = validParameter.validFile(parameters, "nseqs", false); if (temp == "not found"){ temp = "f"; }
199 nseqs = m->isTrue(temp);
201 temp = validParameter.validFile(parameters, "permute", false); if (temp == "not found"){ temp = "f"; }
202 perm = m->isTrue(temp);
204 temp = validParameter.validFile(parameters, "fontsize", false); if (temp == "not found") { temp = "24"; }
205 m->mothurConvert(temp, fontsize);
210 catch(exception& e) {
211 m->errorOut(e, "VennCommand", "VennCommand");
215 //**********************************************************************************************************************
217 int VennCommand::execute(){
220 if (abort == true) { if (calledHelp) { return 0; } return 2; }
222 ValidCalculators validCalculator;
224 if (format == "list") {
225 for (int i=0; i<Estimators.size(); i++) {
226 if (validCalculator.isValidCalculator("vennsingle", Estimators[i]) == true) {
227 if (Estimators[i] == "sobs") {
228 vennCalculators.push_back(new Sobs());
229 }else if (Estimators[i] == "chao") {
230 vennCalculators.push_back(new Chao1());
231 }else if (Estimators[i] == "ace") {
234 vennCalculators.push_back(new Ace(abund));
239 for (int i=0; i<Estimators.size(); i++) {
240 if (validCalculator.isValidCalculator("vennshared", Estimators[i]) == true) {
241 if (Estimators[i] == "sharedsobs") {
242 vennCalculators.push_back(new SharedSobsCS());
243 }else if (Estimators[i] == "sharedchao") {
244 vennCalculators.push_back(new SharedChao1());
245 }else if (Estimators[i] == "sharedace") {
246 vennCalculators.push_back(new SharedAce());
252 //if the users entered no valid calculators don't execute command
253 if (vennCalculators.size() == 0) { m->mothurOut("No valid calculators given, please correct."); m->mothurOutEndLine(); return 0; }
255 venn = new Venn(outputDir, nseqs, inputfile, fontsize);
256 input = new InputData(inputfile, format);
260 if (format == "sharedfile") {
261 lookup = input->getSharedRAbundVectors();
262 lastLabel = lookup[0]->getLabel();
264 if ((lookup.size() > 4) && (perm)) { combosOfFour = findCombinations(lookup.size()); }
265 }else if (format == "list") {
266 sabund = input->getSAbundVector();
267 lastLabel = sabund->getLabel();
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 m->clearGroups(); delete venn; delete input;
283 for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); }
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 m->clearGroups(); delete venn; delete input;
366 for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); }
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
424 if (m->control_pressed) {
425 m->clearGroups(); delete venn; delete input;
426 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
427 for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); }
434 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
436 if (m->control_pressed) {
437 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
438 delete sabund; delete venn; delete input;
439 for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); }
443 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
445 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
446 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
447 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
450 processedLabels.insert(sabund->getLabel());
451 userLabels.erase(sabund->getLabel());
454 if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
455 string saveLabel = sabund->getLabel();
458 sabund = input->getSAbundVector(lastLabel);
460 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
461 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
462 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
465 processedLabels.insert(sabund->getLabel());
466 userLabels.erase(sabund->getLabel());
468 //restore real lastlabel to save below
469 sabund->setLabel(saveLabel);
472 lastLabel = sabund->getLabel();
475 sabund = input->getSAbundVector();
478 if (m->control_pressed) {
479 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
480 for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); }
481 delete venn; delete input;
485 //output error messages about any remaining user labels
486 set<string>::iterator it;
487 bool needToRun = false;
488 for (it = userLabels.begin(); it != userLabels.end(); it++) {
489 m->mothurOut("Your file does not include the label " + *it);
490 if (processedLabels.count(lastLabel) != 1) {
491 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
494 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
498 //run last label if you need to
499 if (needToRun == true) {
500 if (sabund != NULL) { delete sabund; }
501 sabund = input->getSAbundVector(lastLabel);
503 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
504 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
505 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]); } }
511 if (m->control_pressed) {
512 delete venn; delete input;
513 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
514 for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); }
519 for (int i = 0; i < vennCalculators.size(); i++) { delete vennCalculators[i]; }
520 delete venn; delete input;
522 m->mothurOutEndLine();
523 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
524 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
525 m->mothurOutEndLine();
530 catch(exception& e) {
531 m->errorOut(e, "VennCommand", "execute");
535 //**********************************************************************************************************************
536 //returns a vector of sets containing the 4 group combinations
537 set< set<int> > VennCommand::findCombinations(int lookupSize){
539 set< set<int> > combos;
542 for (int i = 0; i < lookupSize; i++) { possibles.insert(i); }
544 getCombos(possibles, combos);
549 catch(exception& e) {
550 m->errorOut(e, "VennCommand", "findCombinations");
554 //**********************************************************************************************************************
555 //recusively finds combos of 4
556 int VennCommand::getCombos(set<int> possibles, set< set<int> >& combos){
559 if (possibles.size() == 4) { //done
560 if (combos.count(possibles) == 0) { //no dups
561 combos.insert(possibles);
563 }else { //we still have work to do
564 set<int>::iterator it;
565 set<int>::iterator it2;
566 for (it = possibles.begin(); it != possibles.end(); it++) {
568 set<int> newPossibles;
569 for (it2 = possibles.begin(); it2 != possibles.end(); it2++) { //all possible combos of one length smaller
571 newPossibles.insert(*it2);
574 getCombos(newPossibles, combos);
580 catch(exception& e) {
581 m->errorOut(e, "VennCommand", "getCombos");
586 //**********************************************************************************************************************