]> git.donarmstrong.com Git - mothur.git/blob - matrixoutputcommand.cpp
added getCommandInfoCommand for gui
[mothur.git] / matrixoutputcommand.cpp
1 /*
2  *  matrixoutputcommand.cpp
3  *  Mothur
4  *
5  *  Created by Sarah Westcott on 5/20/09.
6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
7  *
8  */
9
10 #include "matrixoutputcommand.h"
11 #include "sharedsobscollectsummary.h"
12 #include "sharedchao1.h"
13 #include "sharedace.h"
14 #include "sharednseqs.h"
15 #include "sharedjabund.h"
16 #include "sharedsorabund.h"
17 #include "sharedjclass.h"
18 #include "sharedsorclass.h"
19 #include "sharedjest.h"
20 #include "sharedsorest.h"
21 #include "sharedthetayc.h"
22 #include "sharedthetan.h"
23 #include "sharedkstest.h"
24 #include "whittaker.h"
25 #include "sharedochiai.h"
26 #include "sharedanderbergs.h"
27 #include "sharedkulczynski.h"
28 #include "sharedkulczynskicody.h"
29 #include "sharedlennon.h"
30 #include "sharedmorisitahorn.h"
31 #include "sharedbraycurtis.h"
32 #include "sharedjackknife.h"
33 #include "whittaker.h"
34 #include "odum.h"
35 #include "canberra.h"
36 #include "structeuclidean.h"
37 #include "structchord.h"
38 #include "hellinger.h"
39 #include "manhattan.h"
40 #include "structpearson.h"
41 #include "soergel.h"
42 #include "spearman.h"
43 #include "structkulczynski.h"
44 #include "structchi2.h"
45 #include "speciesprofile.h"
46 #include "hamming.h"
47 #include "gower.h"
48 #include "memchi2.h"
49 #include "memchord.h"
50 #include "memeuclidean.h"
51 #include "mempearson.h"
52
53 //**********************************************************************************************************************
54 vector<string> MatrixOutputCommand::setParameters(){    
55         try {
56                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pshared);
57                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
58                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
59                 CommandParameter pcalc("calc", "Multiple", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-whittaker-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-hamming-structchi2-gower-memchi2-memchord-memeuclidean-mempearson", "jclass-thetayc", "", "", "",true,false); parameters.push_back(pcalc);
60                 CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "",false,false); parameters.push_back(poutput);
61                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
62                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
63                 
64                 vector<string> myArray;
65                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
66                 return myArray;
67         }
68         catch(exception& e) {
69                 m->errorOut(e, "MatrixOutputCommand", "setParameters");
70                 exit(1);
71         }
72 }
73 //**********************************************************************************************************************
74 string MatrixOutputCommand::getHelpString(){    
75         try {
76                 string helpString = "";
77                 ValidCalculators validCalculator;
78                 helpString += "The dist.shared command parameters are shared, groups, calc, output and label.  shared is a required, unless you have a valid current file.\n";
79                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n";
80                 helpString += "The group names are separated by dashes. The label parameter allows you to select what distance levels you would like distance matrices created for, and is also separated by dashes.\n";
81                 helpString += "The dist.shared command should be in the following format: dist.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n";
82                 helpString += "The output parameter allows you to specify format of your distance matrix. Options are lt, and square. The default is lt.\n";
83                 helpString += "Example dist.shared(groups=A-B-C, calc=jabund-sorabund).\n";
84                 helpString += "The default value for groups is all the groups in your groupfile.\n";
85                 helpString += "The default value for calc is jclass and thetayc.\n";
86                 helpString += validCalculator.printCalc("matrix");
87                 helpString += "The dist.shared command outputs a .dist file for each calculator you specify at each distance you choose.\n";
88                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
89                 return helpString;
90         }
91         catch(exception& e) {
92                 m->errorOut(e, "MatrixOutputCommand", "getHelpString");
93                 exit(1);
94         }
95 }
96 //**********************************************************************************************************************
97 MatrixOutputCommand::MatrixOutputCommand(){     
98         try {
99                 abort = true; calledHelp = true; 
100                 setParameters();
101                 vector<string> tempOutNames;
102                 outputTypes["phylip"] = tempOutNames;
103         }
104         catch(exception& e) {
105                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
106                 exit(1);
107         }
108 }
109 //**********************************************************************************************************************
110
111 MatrixOutputCommand::MatrixOutputCommand(string option)  {
112         try {
113                 abort = false; calledHelp = false;   
114                 allLines = 1;
115                                 
116                 //allow user to run help
117                 if(option == "help") {  help(); abort = true; calledHelp = true; }
118                 
119                 else {
120                         vector<string> myArray = setParameters();
121                         
122                         OptionParser parser(option);
123                         map<string,string> parameters  = parser.getParameters();
124                         map<string,string>::iterator it;
125                         
126                         ValidParameters validParameter;
127                 
128                         //check to make sure all parameters are valid for command
129                         for (it = parameters.begin(); it != parameters.end(); it++) { 
130                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
131                         }
132                         
133                         //initialize outputTypes
134                         vector<string> tempOutNames;
135                         outputTypes["phylip"] = tempOutNames;
136                         
137                         //if the user changes the input directory command factory will send this info to us in the output parameter 
138                         string inputDir = validParameter.validFile(parameters, "inputdir", false);              
139                         if (inputDir == "not found"){   inputDir = "";          }
140                         else {
141                                 string path;
142                                 it = parameters.find("shared");
143                                 //user has given a template file
144                                 if(it != parameters.end()){ 
145                                         path = m->hasPath(it->second);
146                                         //if the user has not given a path then, add inputdir. else leave path alone.
147                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
148                                 }
149                         }
150                         
151                         sharedfile = validParameter.validFile(parameters, "shared", true);
152                         if (sharedfile == "not found") {                        
153                                 //if there is a current shared file, use it
154                                 sharedfile = m->getSharedFile(); 
155                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
156                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
157                         }else if (sharedfile == "not open") { sharedfile = ""; abort = true; }
158                         
159                         //if the user changes the output directory command factory will send this info to us in the output parameter 
160                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  
161                                 outputDir = ""; 
162                                 outputDir += m->hasPath(sharedfile); //if user entered a file with a path then preserve it      
163                         }
164                         
165                         //check for optional parameter and set defaults
166                         // ...at some point should added some additional type checking...
167                         label = validParameter.validFile(parameters, "label", false);                   
168                         if (label == "not found") { label = ""; }
169                         else { 
170                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
171                                 else { allLines = 1;  }
172                         }
173                         
174                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "lt"; }
175                         if ((output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are lt and square. I will use lt."); m->mothurOutEndLine(); output = "lt"; }
176                         
177                         groups = validParameter.validFile(parameters, "groups", false);                 
178                         if (groups == "not found") { groups = ""; }
179                         else { 
180                                 m->splitAtDash(groups, Groups);
181                                 m->Groups = Groups;
182                         }
183                                 
184                         calc = validParameter.validFile(parameters, "calc", false);                     
185                         if (calc == "not found") { calc = "jclass-thetayc";  }
186                         else { 
187                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
188                         }
189                         m->splitAtDash(calc, Estimators);
190
191                         if (abort == false) {
192                         
193                                 ValidCalculators validCalculator;
194                                 
195                                 int i;
196                                 for (i=0; i<Estimators.size(); i++) {
197                                         if (validCalculator.isValidCalculator("matrix", Estimators[i]) == true) { 
198                                                 if (Estimators[i] == "sharedsobs") { 
199                                                         matrixCalculators.push_back(new SharedSobsCS());
200                                                 }else if (Estimators[i] == "sharedchao") { 
201                                                         matrixCalculators.push_back(new SharedChao1());
202                                                 }else if (Estimators[i] == "sharedace") { 
203                                                         matrixCalculators.push_back(new SharedAce());
204                                                 }else if (Estimators[i] == "jabund") {  
205                                                         matrixCalculators.push_back(new JAbund());
206                                                 }else if (Estimators[i] == "sorabund") { 
207                                                         matrixCalculators.push_back(new SorAbund());
208                                                 }else if (Estimators[i] == "jclass") { 
209                                                         matrixCalculators.push_back(new Jclass());
210                                                 }else if (Estimators[i] == "sorclass") { 
211                                                         matrixCalculators.push_back(new SorClass());
212                                                 }else if (Estimators[i] == "jest") { 
213                                                         matrixCalculators.push_back(new Jest());
214                                                 }else if (Estimators[i] == "sorest") { 
215                                                         matrixCalculators.push_back(new SorEst());
216                                                 }else if (Estimators[i] == "thetayc") { 
217                                                         matrixCalculators.push_back(new ThetaYC());
218                                                 }else if (Estimators[i] == "thetan") { 
219                                                         matrixCalculators.push_back(new ThetaN());
220                                                 }else if (Estimators[i] == "kstest") { 
221                                                         matrixCalculators.push_back(new KSTest());
222                                                 }else if (Estimators[i] == "sharednseqs") { 
223                                                         matrixCalculators.push_back(new SharedNSeqs());
224                                                 }else if (Estimators[i] == "ochiai") { 
225                                                         matrixCalculators.push_back(new Ochiai());
226                                                 }else if (Estimators[i] == "anderberg") { 
227                                                         matrixCalculators.push_back(new Anderberg());
228                                                 }else if (Estimators[i] == "kulczynski") { 
229                                                         matrixCalculators.push_back(new Kulczynski());
230                                                 }else if (Estimators[i] == "kulczynskicody") { 
231                                                         matrixCalculators.push_back(new KulczynskiCody());
232                                                 }else if (Estimators[i] == "lennon") { 
233                                                         matrixCalculators.push_back(new Lennon());
234                                                 }else if (Estimators[i] == "morisitahorn") { 
235                                                         matrixCalculators.push_back(new MorHorn());
236                                                 }else if (Estimators[i] == "braycurtis") { 
237                                                         matrixCalculators.push_back(new BrayCurtis());
238                                                 }else if (Estimators[i] == "whittaker") { 
239                                                         matrixCalculators.push_back(new Whittaker());
240                                                 }else if (Estimators[i] == "odum") { 
241                                                         matrixCalculators.push_back(new Odum());
242                                                 }else if (Estimators[i] == "canberra") { 
243                                                         matrixCalculators.push_back(new Canberra());
244                                                 }else if (Estimators[i] == "structeuclidean") { 
245                                                         matrixCalculators.push_back(new StructEuclidean());
246                                                 }else if (Estimators[i] == "structchord") { 
247                                                         matrixCalculators.push_back(new StructChord());
248                                                 }else if (Estimators[i] == "hellinger") { 
249                                                         matrixCalculators.push_back(new Hellinger());
250                                                 }else if (Estimators[i] == "manhattan") { 
251                                                         matrixCalculators.push_back(new Manhattan());
252                                                 }else if (Estimators[i] == "structpearson") { 
253                                                         matrixCalculators.push_back(new StructPearson());
254                                                 }else if (Estimators[i] == "soergel") { 
255                                                         matrixCalculators.push_back(new Soergel());
256                                                 }else if (Estimators[i] == "spearman") { 
257                                                         matrixCalculators.push_back(new Spearman());
258                                                 }else if (Estimators[i] == "structkulczynski") { 
259                                                         matrixCalculators.push_back(new StructKulczynski());
260                                                 }else if (Estimators[i] == "speciesprofile") { 
261                                                         matrixCalculators.push_back(new SpeciesProfile());
262                                                 }else if (Estimators[i] == "hamming") { 
263                                                         matrixCalculators.push_back(new Hamming());
264                                                 }else if (Estimators[i] == "structchi2") { 
265                                                         matrixCalculators.push_back(new StructChi2());
266                                                 }else if (Estimators[i] == "gower") { 
267                                                         matrixCalculators.push_back(new Gower());
268                                                 }else if (Estimators[i] == "memchi2") { 
269                                                         matrixCalculators.push_back(new MemChi2());
270                                                 }else if (Estimators[i] == "memchord") { 
271                                                         matrixCalculators.push_back(new MemChord());
272                                                 }else if (Estimators[i] == "memeuclidean") { 
273                                                         matrixCalculators.push_back(new MemEuclidean());
274                                                 }else if (Estimators[i] == "mempearson") { 
275                                                         matrixCalculators.push_back(new MemPearson());
276                                                 }
277                                         }
278                                 }
279                                 
280                         }
281                 }
282                 
283         }
284         catch(exception& e) {
285                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
286                 exit(1);
287         }
288 }
289
290 //**********************************************************************************************************************
291
292 MatrixOutputCommand::~MatrixOutputCommand(){}
293
294 //**********************************************************************************************************************
295
296 int MatrixOutputCommand::execute(){
297         try {
298                 
299                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
300                         
301                 //if the users entered no valid calculators don't execute command
302                 if (matrixCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine();  return 0; }
303                         
304                 input = new InputData(sharedfile, "sharedfile");
305                 lookup = input->getSharedRAbundVectors();
306                 string lastLabel = lookup[0]->getLabel();
307                 
308                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
309                 set<string> processedLabels;
310                 set<string> userLabels = labels;
311                                         
312                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } return 0;}
313                 
314                 numGroups = lookup.size();
315                 
316                 if (m->control_pressed) { delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->Groups.clear(); return 0;  }
317                                 
318                 //as long as you are not at the end of the file or done wih the lines you want
319                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
320                 
321                         if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } m->Groups.clear(); return 0;  }
322                 
323                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){                  
324                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
325                                 process(lookup);
326                                 
327                                 processedLabels.insert(lookup[0]->getLabel());
328                                 userLabels.erase(lookup[0]->getLabel());
329                         }
330                         
331                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
332                                 string saveLabel = lookup[0]->getLabel();
333                                 
334                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } 
335                                 lookup = input->getSharedRAbundVectors(lastLabel);
336
337                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
338                                 process(lookup);
339                                 
340                                 processedLabels.insert(lookup[0]->getLabel());
341                                 userLabels.erase(lookup[0]->getLabel());
342                                 
343                                 //restore real lastlabel to save below
344                                 lookup[0]->setLabel(saveLabel);
345                         }
346
347                         lastLabel = lookup[0]->getLabel();                      
348                         
349                         //get next line to process
350                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } 
351                         lookup = input->getSharedRAbundVectors();
352                 }
353                 
354                 if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } m->Groups.clear(); return 0;  }
355
356                 //output error messages about any remaining user labels
357                 set<string>::iterator it;
358                 bool needToRun = false;
359                 for (it = userLabels.begin(); it != userLabels.end(); it++) {  
360                         m->mothurOut("Your file does not include the label " + *it);  
361                         if (processedLabels.count(lastLabel) != 1) {
362                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
363                                 needToRun = true;
364                         }else {
365                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
366                         }
367                 }
368                 
369                 if (m->control_pressed) { outputTypes.clear(); delete input;  for (int i = 0; i < outputNames.size(); i++) {    remove(outputNames[i].c_str()); } m->Groups.clear(); return 0;  }
370
371                 //run last label if you need to
372                 if (needToRun == true)  {
373                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {  delete lookup[i]; }  } 
374                         lookup = input->getSharedRAbundVectors(lastLabel);
375
376                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
377                         process(lookup);
378                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } 
379                 }
380                 
381                 if (m->control_pressed) { outputTypes.clear();  delete input;  for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str()); } m->Groups.clear(); return 0;  }
382                 
383                 //reset groups parameter
384                 m->Groups.clear();  
385                 
386                 //set phylip file as new current phylipfile
387                 string current = "";
388                 itTypes = outputTypes.find("phylip");
389                 if (itTypes != outputTypes.end()) {
390                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
391                 }
392                 
393                 m->mothurOutEndLine();
394                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
395                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
396                 m->mothurOutEndLine();
397
398
399                 return 0;
400         }
401         catch(exception& e) {
402                 m->errorOut(e, "MatrixOutputCommand", "execute");
403                 exit(1);
404         }
405 }
406 /***********************************************************/
407 void MatrixOutputCommand::printSims(ostream& out) {
408         try {
409                 
410                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
411                 
412                 //output num seqs
413                 out << simMatrix.size() << endl;
414                 
415                 if (output == "lt") {
416                         for (int m = 0; m < simMatrix.size(); m++)      {
417                                 out << lookup[m]->getGroup() << '\t';
418                                 for (int n = 0; n < m; n++)     {
419                                         out << simMatrix[m][n] << '\t'; 
420                                 }
421                                 out << endl;
422                         }
423                 }else{
424                         for (int m = 0; m < simMatrix.size(); m++)      {
425                                 out << lookup[m]->getGroup() << '\t';
426                                 for (int n = 0; n < simMatrix[m].size(); n++)   {
427                                         out << simMatrix[m][n] << '\t'; 
428                                 }
429                                 out << endl;
430                         }
431                 }
432         }
433         catch(exception& e) {
434                 m->errorOut(e, "MatrixOutputCommand", "printSims");
435                 exit(1);
436         }
437 }
438 /***********************************************************/
439 int MatrixOutputCommand::process(vector<SharedRAbundVector*> thisLookup){
440         try {
441         
442                                 EstOutput data;
443                                 vector<SharedRAbundVector*> subset;
444
445                                 //for each calculator                                                                                           
446                                 for(int i = 0 ; i < matrixCalculators.size(); i++) {
447                                         
448                                         //initialize simMatrix
449                                         simMatrix.clear();
450                                         simMatrix.resize(numGroups);
451                                         for (int p = 0; p < simMatrix.size(); p++)      {
452                                                 for (int j = 0; j < simMatrix.size(); j++)      {
453                                                         simMatrix[p].push_back(0.0);
454                                                 }
455                                         }
456                                 
457                                         for (int k = 0; k < thisLookup.size(); k++) { 
458                                                 for (int l = k; l < thisLookup.size(); l++) {
459                                                         if (k != l) { //we dont need to similiarity of a groups to itself
460                                                                 //get estimated similarity between 2 groups
461                                                                 
462                                                                 if (m->control_pressed) { return 0; }
463                                                                 
464                                                                 subset.clear(); //clear out old pair of sharedrabunds
465                                                                 //add new pair of sharedrabunds
466                                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]); 
467                                                                 
468                                                                 //if this calc needs all groups to calculate the pair load all groups
469                                                                 if (matrixCalculators[i]->getNeedsAll()) { 
470                                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
471                                                                         for (int w = 0; w < thisLookup.size(); w++) {
472                                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
473                                                                         }
474                                                                 }
475                                                                 
476                                                                 data = matrixCalculators[i]->getValues(subset); //saves the calculator outputs
477                                                                 //save values in similarity matrix
478                                                                 simMatrix[k][l] = 1.0 - data[0];  //convert similiarity to distance
479                                                                 simMatrix[l][k] = 1.0 - data[0];  //convert similiarity to distance
480                                                         }
481                                                 }
482                                         }
483                                         
484                                         exportFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".dist";
485                                         m->openOutputFile(exportFileName, out);
486                                         outputNames.push_back(exportFileName); outputTypes["phylip"].push_back(exportFileName);
487                                         
488                                         printSims(out);
489                                         out.close();
490                                         
491                                 }
492
493                                 return 0;
494                 
495         }
496         catch(exception& e) {
497                 m->errorOut(e, "MatrixOutputCommand", "process");
498                 exit(1);
499         }
500 }
501 /***********************************************************/
502
503
504