]> git.donarmstrong.com Git - mothur.git/blob - heatmapsimcommand.cpp
working on citations
[mothur.git] / heatmapsimcommand.cpp
1 /*
2  *  heatmapsimcommand.cpp
3  *  Mothur
4  *
5  *  Created by Sarah Westcott on 6/8/09.
6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
7  *
8  */
9
10 #include "heatmapsimcommand.h"
11 #include "sharedjabund.h"
12 #include "sharedsorabund.h"
13 #include "sharedjclass.h"
14 #include "sharedsorclass.h"
15 #include "sharedjest.h"
16 #include "sharedsorest.h"
17 #include "sharedthetayc.h"
18 #include "sharedthetan.h"
19 #include "sharedmorisitahorn.h"
20 #include "sharedbraycurtis.h"
21
22
23 //**********************************************************************************************************************
24 vector<string> HeatMapSimCommand::setParameters(){      
25         try {
26                 CommandParameter pshared("shared", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pshared);        
27                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pphylip);
28                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName",false,false); parameters.push_back(pname);
29                 CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "ColumnName",false,false); parameters.push_back(pcolumn);          
30                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
31                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
32                 CommandParameter pcalc("calc", "Multiple", "jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-morisitahorn-braycurtis", "jest-thetayc", "", "", "",true,false); parameters.push_back(pcalc);
33                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
34                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
35                 
36                 vector<string> myArray;
37                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
38                 return myArray;
39         }
40         catch(exception& e) {
41                 m->errorOut(e, "HeatMapSimCommand", "setParameters");
42                 exit(1);
43         }
44 }
45 //**********************************************************************************************************************
46 string HeatMapSimCommand::getHelpString(){      
47         try {
48                 string helpString = "";
49                 ValidCalculators validCalculator;
50                 helpString += "The heatmap.sim command parameters are shared, phylip, column, name, groups, calc and label.  shared or phylip or column and name are required unless valid current files exist.\n";
51                 helpString += "There are two ways to use the heatmap.sim command. The first is with the read.otu command. \n";
52                 helpString += "With the read.otu command you may use the groups, label and calc parameters. \n";
53                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included in your heatmap.\n";
54                 helpString += "The group names are separated by dashes. The label parameter allows you to select what distance levels you would like a heatmap created for, and is also separated by dashes.\n";
55                 helpString += "The heatmap.sim command should be in the following format: heatmap.sim(groups=yourGroups, calc=yourCalc, label=yourLabels).\n";
56                 helpString += "Example heatmap.sim(groups=A-B-C, calc=jabund).\n";
57                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
58                 helpString +=  validCalculator.printCalc("heat");
59                 helpString += "The default value for calc is jclass-thetayc.\n";
60                 helpString += "The heatmap.sim command outputs a .svg file for each calculator you choose at each label you specify.\n";
61                 helpString += "The second way to use the heatmap.sim command is with a distance file representing the distance bewteen your groups. \n";
62                 helpString += "Using the command this way, the phylip or column parameter are required, and only one may be used.  If you use a column file the name filename is required. \n";
63                 helpString += "The heatmap.sim command should be in the following format: heatmap.sim(phylip=yourDistanceFile).\n";
64                 helpString += "Example heatmap.sim(phylip=amazonGroups.dist).\n";
65                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
66                 return helpString;
67         }
68         catch(exception& e) {
69                 m->errorOut(e, "HeatMapSimCommand", "getHelpString");
70                 exit(1);
71         }
72 }
73 //**********************************************************************************************************************
74 HeatMapSimCommand::HeatMapSimCommand(){ 
75         try {
76                 abort = true; calledHelp = true; 
77                 setParameters();
78                 vector<string> tempOutNames;
79                 outputTypes["svg"] = tempOutNames;
80         }
81         catch(exception& e) {
82                 m->errorOut(e, "HeatMapSimCommand", "HeatMapSimCommand");
83                 exit(1);
84         }
85 }
86 //**********************************************************************************************************************
87
88 HeatMapSimCommand::HeatMapSimCommand(string option)  {
89         try {
90                 abort = false; calledHelp = false;   
91                 allLines = 1;
92                         
93                 //allow user to run help
94                 if(option == "help") {  help(); abort = true; calledHelp = true; }
95                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
96                 
97                 else {
98                         vector<string> myArray = setParameters();
99                         
100                         OptionParser parser(option);
101                         map<string,string> parameters = parser.getParameters();
102                         
103                         ValidParameters validParameter;
104                         map<string,string>::iterator it;
105                         
106                         //check to make sure all parameters are valid for command
107                         for (it = parameters.begin(); it != parameters.end(); it++) { 
108                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
109                         }
110                         
111                         //initialize outputTypes
112                         vector<string> tempOutNames;
113                         outputTypes["svg"] = tempOutNames;
114                         
115                         format = "";
116                         //if the user changes the output directory command factory will send this info to us in the output parameter 
117                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
118                         
119                         //if the user changes the input directory command factory will send this info to us in the output parameter 
120                         string inputDir = validParameter.validFile(parameters, "inputdir", false);              
121                         if (inputDir == "not found"){   inputDir = "";          }
122                         else {
123                                 string path;
124                                 it = parameters.find("phylip");
125                                 //user has given a template file
126                                 if(it != parameters.end()){ 
127                                         path = m->hasPath(it->second);
128                                         //if the user has not given a path then, add inputdir. else leave path alone.
129                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
130                                 }
131                                 
132                                 it = parameters.find("column");
133                                 //user has given a template file
134                                 if(it != parameters.end()){ 
135                                         path = m->hasPath(it->second);
136                                         //if the user has not given a path then, add inputdir. else leave path alone.
137                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
138                                 }
139                                 
140                                 it = parameters.find("name");
141                                 //user has given a template file
142                                 if(it != parameters.end()){ 
143                                         path = m->hasPath(it->second);
144                                         //if the user has not given a path then, add inputdir. else leave path alone.
145                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
146                                 }
147                                 
148                                 it = parameters.find("shared");
149                                 //user has given a template file
150                                 if(it != parameters.end()){ 
151                                         path = m->hasPath(it->second);
152                                         //if the user has not given a path then, add inputdir. else leave path alone.
153                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
154                                 }
155                         }
156
157                         //required parameters
158                         phylipfile = validParameter.validFile(parameters, "phylip", true);
159                         if (phylipfile == "not open") { abort = true; }
160                         else if (phylipfile == "not found") { phylipfile = ""; }        
161                         else {  format = "phylip";      inputfile = phylipfile; if (outputDir == "") { outputDir += m->hasPath(phylipfile); }  }
162                         
163                         columnfile = validParameter.validFile(parameters, "column", true);
164                         if (columnfile == "not open") { abort = true; } 
165                         else if (columnfile == "not found") { columnfile = ""; }
166                         else {  format = "column";      inputfile = columnfile; if (outputDir == "") { outputDir += m->hasPath(columnfile); } }
167                         
168                         sharedfile = validParameter.validFile(parameters, "shared", true);
169                         if (sharedfile == "not open") { abort = true; } 
170                         else if (sharedfile == "not found") { sharedfile = ""; }
171                         else {  format = "shared";      inputfile = sharedfile;  if (outputDir == "") { outputDir += m->hasPath(sharedfile); } }
172                         
173                         namefile = validParameter.validFile(parameters, "name", true);
174                         if (namefile == "not open") { abort = true; }   
175                         else if (namefile == "not found") { namefile = ""; }
176                         
177                         
178                         //error checking on files                       
179                         if ((sharedfile == "") && ((phylipfile == "") && (columnfile == "")))   { 
180                                 sharedfile = m->getSharedFile(); 
181                                 if (sharedfile != "") { format = "shared"; inputfile = sharedfile; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
182                                 else { 
183                                         //is there are current file available for either of these?
184                                         //give priority to column, then phylip
185                                         columnfile = m->getColumnFile(); 
186                                         if (columnfile != "") {  format = "column"; inputfile = columnfile; m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
187                                         else { 
188                                                 phylipfile = m->getPhylipFile(); 
189                                                 if (phylipfile != "") { format = "phylip";  inputfile = phylipfile; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
190                                                 else { 
191                                                         m->mothurOut("No valid current files. You must provide a shared or phylip or column file."); m->mothurOutEndLine(); 
192                                                         abort = true;
193                                                 }
194                                         }
195                                 }
196                         }
197                         else if ((phylipfile != "") && (columnfile != "")) { m->mothurOut("When running the heatmap.sim command with a distance file you may not use both the column and the phylip parameters."); m->mothurOutEndLine(); abort = true; }
198                         
199                         if (columnfile != "") {
200                                 if (namefile == "") { 
201                                         namefile = m->getNameFile(); 
202                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
203                                         else { 
204                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine(); 
205                                                 abort = true; 
206                                         }       
207                                 }
208                         }
209                         
210                         //check for optional parameter and set defaults
211                         // ...at some point should added some additional type checking...
212                                                         
213                         label = validParameter.validFile(parameters, "label", false);                   
214                         if (label == "not found") { label = ""; }
215                         else { 
216                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
217                                 else { allLines = 1;  }
218                         }
219                                 
220                         calc = validParameter.validFile(parameters, "calc", false);                     
221                         if (calc == "not found") { calc = "jest-thetayc";  }
222                         else { 
223                                 if (calc == "default")  {  calc = "jest-thetayc";  }
224                         }
225                         m->splitAtDash(calc, Estimators);
226                         if (m->inUsersGroups("citation", Estimators)) { 
227                                 ValidCalculators validCalc; validCalc.printCitations(Estimators); 
228                                 //remove citation from list of calcs
229                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
230                         }
231                                 
232                         groups = validParameter.validFile(parameters, "groups", false);                 
233                         if (groups == "not found") { groups = ""; }
234                         else { 
235                                 m->splitAtDash(groups, Groups);
236                                 m->Groups = Groups;
237                         }
238                         
239                         
240                         if (abort == false) {
241                                 ValidCalculators validCalculator;
242                         
243                                 int i;
244                                 for (i=0; i<Estimators.size(); i++) {
245                                         if (validCalculator.isValidCalculator("heat", Estimators[i]) == true) { 
246                                                 if (Estimators[i] == "jabund") {        
247                                                         heatCalculators.push_back(new JAbund());
248                                                 }else if (Estimators[i] == "sorabund") { 
249                                                         heatCalculators.push_back(new SorAbund());
250                                                 }else if (Estimators[i] == "jclass") { 
251                                                         heatCalculators.push_back(new Jclass());
252                                                 }else if (Estimators[i] == "sorclass") { 
253                                                         heatCalculators.push_back(new SorClass());
254                                                 }else if (Estimators[i] == "jest") { 
255                                                         heatCalculators.push_back(new Jest());
256                                                 }else if (Estimators[i] == "sorest") { 
257                                                         heatCalculators.push_back(new SorEst());
258                                                 }else if (Estimators[i] == "thetayc") { 
259                                                         heatCalculators.push_back(new ThetaYC());
260                                                 }else if (Estimators[i] == "thetan") { 
261                                                         heatCalculators.push_back(new ThetaN());
262                                                 }else if (Estimators[i] == "morisitahorn") { 
263                                                         heatCalculators.push_back(new MorHorn());
264                                                 }else if (Estimators[i] == "braycurtis") { 
265                                                         heatCalculators.push_back(new BrayCurtis());
266                                                 }
267                                         }
268                                 }
269                                 
270                         }
271                 }
272
273                                 
274
275         }
276         catch(exception& e) {
277                 m->errorOut(e, "HeatMapSimCommand", "HeatMapSimCommand");
278                 exit(1);
279         }
280 }
281 //**********************************************************************************************************************
282
283 int HeatMapSimCommand::execute(){
284         try {
285         
286                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
287                 
288                 heatmap = new HeatMapSim(outputDir, inputfile);
289                 
290                 if (format == "shared") {
291                         runCommandShared();
292                 }else{  runCommandDist();       }
293                 
294                 delete heatmap;
295                 
296                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } outputTypes.clear(); return 0; }
297                 
298                 m->mothurOutEndLine();
299                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
300                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
301                 m->mothurOutEndLine();
302                 
303                 return 0;
304         }
305         catch(exception& e) {
306                 m->errorOut(e, "HeatMapSimCommand", "execute");
307                 exit(1);
308         }
309 }
310
311 //**********************************************************************************************************************
312 int HeatMapSimCommand::runCommandShared() {
313         try {
314                 //if the users entered no valid calculators don't execute command
315                 if (heatCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine(); return 0; }
316                 
317                 input = new InputData(sharedfile, "sharedfile");
318                 lookup = input->getSharedRAbundVectors();
319                 string lastLabel = lookup[0]->getLabel();
320                         
321                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0;}
322                                 
323                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
324                 set<string> processedLabels;
325                 set<string> userLabels = labels;
326                 
327                 if (m->control_pressed) {  delete input;  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  m->Groups.clear(); return 0; }
328                 
329                 //as long as you are not at the end of the file or done wih the lines you want
330                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
331                         
332                         if (m->control_pressed) { delete input;  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->Groups.clear(); return 0; }
333
334                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){                  
335         
336                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
337                                 vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
338                                 for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]);  outputTypes["svg"].push_back(outfilenames[i]); }
339                                         
340                                 processedLabels.insert(lookup[0]->getLabel());
341                                 userLabels.erase(lookup[0]->getLabel());
342                         }
343                                 
344                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
345                                 string saveLabel = lookup[0]->getLabel();
346                         
347                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } 
348                                 lookup = input->getSharedRAbundVectors(lastLabel);                              
349
350                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
351                                 vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
352                                 for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]);  }
353                                         
354                                 processedLabels.insert(lookup[0]->getLabel());
355                                 userLabels.erase(lookup[0]->getLabel());
356                                 
357                                 //restore real lastlabel to save below
358                                 lookup[0]->setLabel(saveLabel);
359                         }
360                                 
361                         //prevent memory leak
362                          
363                         lastLabel = lookup[0]->getLabel();                      
364
365                         //get next line to process
366                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } 
367                         lookup = input->getSharedRAbundVectors();
368
369                 }
370                 
371                         
372                 if (m->control_pressed) {  delete input;  m->Groups.clear();  return 0; }
373
374                 //output error messages about any remaining user labels
375                 set<string>::iterator it;
376                 bool needToRun = false;
377                 for (it = userLabels.begin(); it != userLabels.end(); it++) {  
378                         m->mothurOut("Your file does not include the label " + *it); 
379                         if (processedLabels.count(lastLabel) != 1) {
380                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
381                                 needToRun = true;
382                         }else {
383                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
384                         }
385                 }
386                 
387                 if (m->control_pressed) {  delete input;  m->Groups.clear(); return 0; }
388                 
389                 //run last label if you need to
390                 if (needToRun == true)  {
391                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) { delete lookup[i]; } } 
392                         lookup = input->getSharedRAbundVectors(lastLabel);                              
393
394                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
395                         vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
396                         for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]);  }
397                         
398                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } 
399                 }
400                 
401                 if (m->control_pressed) {  delete input;  m->Groups.clear(); return 0; }
402                         
403                 //reset groups parameter
404                 m->Groups.clear();  
405                         
406                 delete input;  
407         
408                 return 0;
409         }
410         catch(exception& e) {
411                 m->errorOut(e, "HeatMapSimCommand", "runCommandShared");
412                 exit(1);
413         }
414 }
415 //**********************************************************************************************************************
416 int HeatMapSimCommand::runCommandDist() {
417         try {
418         
419                 vector< vector<double> > matrix;
420                 vector<string> names;
421                 ifstream in;
422                 
423                 //read distance file and create distance vector and names vector
424                 if (format == "phylip") {
425                         //read phylip file
426                         m->openInputFile(phylipfile, in);
427                         
428                         string name;
429                         int numSeqs;
430                         in >> numSeqs >> name; 
431                         
432                         //save name
433                         names.push_back(name);
434                 
435                         //resize the matrix and fill with zeros
436                         matrix.resize(numSeqs); 
437                         for(int i = 0; i < numSeqs; i++) {
438                                 matrix[i].resize(numSeqs, 0.0);
439                         }
440                                         
441                         //determine if matrix is square or lower triangle
442                         //if it is square read the distances for the first sequence
443                         char d;
444                         bool square;
445                         while((d=in.get()) != EOF){
446                                 
447                                 //is d a number meaning its square
448                                 if(isalnum(d)){ 
449                                         square = true;
450                                         in.putback(d);
451                                         
452                                         for(int i=0;i<numSeqs;i++){
453                                                 in >> matrix[0][i];
454                                         }
455                                         break;
456                                 }
457                                 
458                                 //is d a line return meaning its lower triangle
459                                 if(d == '\n'){
460                                         square = false;
461                                         break;
462                                 }
463                         }
464                         
465                         //read rest of matrix
466                         if (square == true) { 
467                                 for(int i=1;i<numSeqs;i++){
468                                         in >> name;             
469                                         names.push_back(name);
470                                         
471                                         if (m->control_pressed) { return 0; }
472                                         
473                                         for(int j=0;j<numSeqs;j++) { in >> matrix[i][j];  }
474                                         m->gobble(in);
475                                 }
476                         }else { 
477                                 double dist;
478                                 for(int i=1;i<numSeqs;i++){
479                                         in >> name;     
480                                         names.push_back(name);  
481                                         
482                                         if (m->control_pressed) { return 0; }
483                                         
484                                         for(int j=0;j<i;j++){
485                                                 in >> dist;
486                                                 matrix[i][j] = dist;  matrix[j][i] = dist;
487                                         }
488                                         m->gobble(in);
489                                 }
490                         }
491                         in.close();
492                 }else {
493                         //read names file
494                         NameAssignment* nameMap = new NameAssignment(namefile);
495                         nameMap->readMap();
496                         
497                         //put names in order in vector
498                         for (int i = 0; i < nameMap->size(); i++) {
499                                 names.push_back(nameMap->get(i));
500                         }
501                         
502                         //resize matrix
503                         matrix.resize(nameMap->size());
504                         for (int i = 0; i < nameMap->size(); i++) {
505                                 matrix[i].resize(nameMap->size(), 0.0);
506                         }
507                         
508                         //read column file
509                         string first, second;
510                         double dist;
511                         m->openInputFile(columnfile, in);
512                         
513                         while (!in.eof()) {
514                                 in >> first >> second >> dist; m->gobble(in);
515                                 
516                                 if (m->control_pressed) { return 0; }
517                                 
518                                 map<string, int>::iterator itA = nameMap->find(first);
519                                 map<string, int>::iterator itB = nameMap->find(second);
520                                 
521                                 if(itA == nameMap->end()){  m->mothurOut("AAError: Sequence '" + first + "' was not found in the names file, please correct\n"); exit(1);  }
522                                 if(itB == nameMap->end()){  m->mothurOut("ABError: Sequence '" + second + "' was not found in the names file, please correct\n"); exit(1);  }
523                                 
524                                 //save distance
525                                 matrix[itA->second][itB->second] = dist;
526                                 matrix[itB->second][itA->second] = dist;
527                         }
528                         in.close();
529                         
530                         delete nameMap;
531                 }
532                 
533                 
534                 string outputFileName = heatmap->getPic(matrix, names);
535                 outputNames.push_back(outputFileName); //vector<vector<double>>, vector<string>
536                 outputTypes["svg"].push_back(outputFileName);
537                 
538                 return 0;
539         }
540         catch(exception& e) {
541                 m->errorOut(e, "HeatMapSimCommand", "runCommandDist");
542                 exit(1);
543         }
544 }
545 //**********************************************************************************************************************
546
547
548
549
550
551