]> git.donarmstrong.com Git - mothur.git/blob - getseqscommand.cpp
changes while testing
[mothur.git] / getseqscommand.cpp
1 /*
2  *  getseqscommand.cpp
3  *  Mothur
4  *
5  *  Created by Sarah Westcott on 7/8/09.
6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
7  *
8  */
9
10 #include "getseqscommand.h"
11 #include "sequence.hpp"
12 #include "listvector.hpp"
13 #include "counttable.h"
14
15 //**********************************************************************************************************************
16 vector<string> GetSeqsCommand::setParameters(){ 
17         try {
18                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none","fasta",false,false,true); parameters.push_back(pfasta);
19         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "FNGLT", "none","name",false,false,true); parameters.push_back(pname);
20         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "FNGLT", "none","count",false,false,true); parameters.push_back(pcount);
21                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "FNGLT", "none","group",false,false,true); parameters.push_back(pgroup);
22                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none","list",false,false,true); parameters.push_back(plist);
23                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none","taxonomy",false,false,true); parameters.push_back(ptaxonomy);
24                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none","alignreport",false,false); parameters.push_back(palignreport);
25                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none","qfile",false,false); parameters.push_back(pqfile);
26                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(paccnos);
27                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pdups);
28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
30                 CommandParameter paccnos2("accnos2", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(paccnos2);
31
32                 vector<string> myArray;
33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
34                 return myArray;
35         }
36         catch(exception& e) {
37                 m->errorOut(e, "GetSeqsCommand", "setParameters");
38                 exit(1);
39         }
40 }
41 //**********************************************************************************************************************
42 string GetSeqsCommand::getHelpString(){ 
43         try {
44                 string helpString = "";
45                 helpString += "The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, count, list, taxonomy, quality or alignreport file.\n";
46                 helpString += "It outputs a file containing only the sequences in the .accnos file.\n";
47                 helpString += "The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos unless you have a valid current accnos file, and at least one of the other parameters.\n";
48                 helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=true. \n";
49                 helpString += "The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
50                 helpString += "Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
52                 return helpString;
53         }
54         catch(exception& e) {
55                 m->errorOut(e, "GetSeqsCommand", "getHelpString");
56                 exit(1);
57         }
58 }
59
60 //**********************************************************************************************************************
61 GetSeqsCommand::GetSeqsCommand(){       
62         try {
63                 abort = true; calledHelp = true;
64                 setParameters();
65                 vector<string> tempOutNames;
66                 outputTypes["fasta"] = tempOutNames;
67                 outputTypes["taxonomy"] = tempOutNames;
68                 outputTypes["name"] = tempOutNames;
69                 outputTypes["group"] = tempOutNames;
70                 outputTypes["alignreport"] = tempOutNames;
71                 outputTypes["list"] = tempOutNames;
72                 outputTypes["qfile"] = tempOutNames;
73         outputTypes["count"] = tempOutNames;
74                 outputTypes["accnosreport"] = tempOutNames;
75         }
76         catch(exception& e) {
77                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
78                 exit(1);
79         }
80 }
81 //**********************************************************************************************************************
82 string GetSeqsCommand::getOutputPattern(string type) {
83     try {
84         string pattern = "";
85         
86         if (type == "fasta")            {   pattern = "[filename],pick,[extension]";    }
87         else if (type == "taxonomy")    {   pattern = "[filename],pick,[extension]";    }
88         else if (type == "name")        {   pattern = "[filename],pick,[extension]";    }
89         else if (type == "group")       {   pattern = "[filename],pick,[extension]";    }
90         else if (type == "count")       {   pattern = "[filename],pick,[extension]";    }
91         else if (type == "list")        {   pattern = "[filename],pick,[extension]";    }
92         else if (type == "qfile")       {   pattern = "[filename],pick,[extension]";    }
93         else if (type == "accnosreport")      {   pattern = "[filename],pick.accnos.report";    }
94         else if (type == "alignreport")      {   pattern = "[filename],pick.align.report";    }
95         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
96         
97         return pattern;
98     }
99     catch(exception& e) {
100         m->errorOut(e, "GetSeqsCommand", "getOutputPattern");
101         exit(1);
102     }
103 }
104 //**********************************************************************************************************************
105 GetSeqsCommand::GetSeqsCommand(string option)  {
106         try {
107                 abort = false; calledHelp = false;   
108                                 
109                 //allow user to run help
110                 if(option == "help") { help(); abort = true; calledHelp = true; }
111                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
112                 
113                 else {
114                         vector<string> myArray = setParameters();
115                         
116                         OptionParser parser(option);
117                         map<string,string> parameters = parser.getParameters();
118                         
119                         ValidParameters validParameter;
120                         map<string,string>::iterator it;
121                         
122                         //check to make sure all parameters are valid for command
123                         for (it = parameters.begin(); it != parameters.end(); it++) { 
124                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
125                         }
126                         
127                         //initialize outputTypes
128                         vector<string> tempOutNames;
129                         outputTypes["fasta"] = tempOutNames;
130                         outputTypes["taxonomy"] = tempOutNames;
131                         outputTypes["name"] = tempOutNames;
132                         outputTypes["group"] = tempOutNames;
133                         outputTypes["alignreport"] = tempOutNames;
134                         outputTypes["list"] = tempOutNames;
135                         outputTypes["qfile"] = tempOutNames;
136                         outputTypes["accnosreport"] = tempOutNames;
137             outputTypes["count"] = tempOutNames;
138                         
139                         //if the user changes the output directory command factory will send this info to us in the output parameter 
140                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
141                         
142                         //if the user changes the input directory command factory will send this info to us in the output parameter 
143                         string inputDir = validParameter.validFile(parameters, "inputdir", false);              
144                         if (inputDir == "not found"){   inputDir = "";          }
145                         else {
146                                 string path;
147                                 it = parameters.find("alignreport");
148                                 //user has given a template file
149                                 if(it != parameters.end()){ 
150                                         path = m->hasPath(it->second);
151                                         //if the user has not given a path then, add inputdir. else leave path alone.
152                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
153                                 }
154                                 
155                                 it = parameters.find("fasta");
156                                 //user has given a template file
157                                 if(it != parameters.end()){ 
158                                         path = m->hasPath(it->second);
159                                         //if the user has not given a path then, add inputdir. else leave path alone.
160                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
161                                 }
162                                 
163                                 it = parameters.find("accnos");
164                                 //user has given a template file
165                                 if(it != parameters.end()){ 
166                                         path = m->hasPath(it->second);
167                                         //if the user has not given a path then, add inputdir. else leave path alone.
168                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
169                                 }
170                                 
171                                 it = parameters.find("accnos2");
172                                 //user has given a template file
173                                 if(it != parameters.end()){ 
174                                         path = m->hasPath(it->second);
175                                         //if the user has not given a path then, add inputdir. else leave path alone.
176                                         if (path == "") {       parameters["accnos2"] = inputDir + it->second;          }
177                                 }
178                                 
179                                 it = parameters.find("list");
180                                 //user has given a template file
181                                 if(it != parameters.end()){ 
182                                         path = m->hasPath(it->second);
183                                         //if the user has not given a path then, add inputdir. else leave path alone.
184                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
185                                 }
186                                 
187                                 it = parameters.find("name");
188                                 //user has given a template file
189                                 if(it != parameters.end()){ 
190                                         path = m->hasPath(it->second);
191                                         //if the user has not given a path then, add inputdir. else leave path alone.
192                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
193                                 }
194                                 
195                                 it = parameters.find("group");
196                                 //user has given a template file
197                                 if(it != parameters.end()){ 
198                                         path = m->hasPath(it->second);
199                                         //if the user has not given a path then, add inputdir. else leave path alone.
200                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
201                                 }
202                                 
203                                 it = parameters.find("taxonomy");
204                                 //user has given a template file
205                                 if(it != parameters.end()){ 
206                                         path = m->hasPath(it->second);
207                                         //if the user has not given a path then, add inputdir. else leave path alone.
208                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
209                                 }
210                                 
211                                 it = parameters.find("qfile");
212                                 //user has given a template file
213                                 if(it != parameters.end()){ 
214                                         path = m->hasPath(it->second);
215                                         //if the user has not given a path then, add inputdir. else leave path alone.
216                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
217                                 }
218                 
219                 it = parameters.find("count");
220                                 //user has given a template file
221                                 if(it != parameters.end()){ 
222                                         path = m->hasPath(it->second);
223                                         //if the user has not given a path then, add inputdir. else leave path alone.
224                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
225                                 }
226                         }
227
228                         
229                         //check for required parameters
230                         accnosfile = validParameter.validFile(parameters, "accnos", true);
231                         if (accnosfile == "not open") { abort = true; }
232                         else if (accnosfile == "not found") {  
233                                 accnosfile = m->getAccnosFile(); 
234                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
235                                 else { 
236                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine(); 
237                                         abort = true;
238                                 } 
239                         }else { m->setAccnosFile(accnosfile); } 
240                         
241                         if (accnosfile2 == "not found") { accnosfile2 = ""; }
242                         
243                         fastafile = validParameter.validFile(parameters, "fasta", true);
244                         if (fastafile == "not open") { fastafile = ""; abort = true; }
245                         else if (fastafile == "not found") {  fastafile = "";  }
246                         else { m->setFastaFile(fastafile); }
247                         
248                         namefile = validParameter.validFile(parameters, "name", true);
249                         if (namefile == "not open") { namefile = ""; abort = true; }
250                         else if (namefile == "not found") {  namefile = "";  }  
251                         else { m->setNameFile(namefile); }
252                         
253                         groupfile = validParameter.validFile(parameters, "group", true);
254                         if (groupfile == "not open") { abort = true; }
255                         else if (groupfile == "not found") {  groupfile = "";  }        
256                         else { m->setGroupFile(groupfile); }
257                         
258                         alignfile = validParameter.validFile(parameters, "alignreport", true);
259                         if (alignfile == "not open") { abort = true; }
260                         else if (alignfile == "not found") {  alignfile = "";  }
261                         
262                         listfile = validParameter.validFile(parameters, "list", true);
263                         if (listfile == "not open") { abort = true; }
264                         else if (listfile == "not found") {  listfile = "";  }
265                         else { m->setListFile(listfile); }
266                         
267                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
268                         if (taxfile == "not open") { taxfile = ""; abort = true; }
269                         else if (taxfile == "not found") {  taxfile = "";  }
270                         else { m->setTaxonomyFile(taxfile); }
271                         
272                         qualfile = validParameter.validFile(parameters, "qfile", true);
273                         if (qualfile == "not open") { abort = true; }
274                         else if (qualfile == "not found") {  qualfile = "";  }
275                         else { m->setQualFile(qualfile); }
276                         
277                         accnosfile2 = validParameter.validFile(parameters, "accnos2", true);
278                         if (accnosfile2 == "not open") { abort = true; }
279                         else if (accnosfile2 == "not found") {  accnosfile2 = "";  }
280                         
281             countfile = validParameter.validFile(parameters, "count", true);
282             if (countfile == "not open") { countfile = ""; abort = true; }
283             else if (countfile == "not found") { countfile = "";  }     
284             else { m->setCountTableFile(countfile); }
285             
286             if ((namefile != "") && (countfile != "")) {
287                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
288             }
289             
290             if ((groupfile != "") && (countfile != "")) {
291                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
292             }
293
294                         
295                         string usedDups = "true";
296                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "true"; usedDups = ""; }
297                         dups = m->isTrue(temp);
298                         
299                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == "") && (accnosfile2 == "") && (countfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, count, alignreport, taxonomy, quality or listfile."); m->mothurOutEndLine(); abort = true; }
300             
301             if (countfile == "") {
302                 if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
303                     vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
304                     parser.getNameFile(files);
305                 }
306             }
307                 }
308
309         }
310         catch(exception& e) {
311                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
312                 exit(1);
313         }
314 }
315 //**********************************************************************************************************************
316
317 int GetSeqsCommand::execute(){
318         try {
319                 
320                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
321                 
322                 //get names you want to keep
323                 names = m->readAccnos(accnosfile);
324                 
325                 if (m->control_pressed) { return 0; }
326         
327         if (countfile != "") {
328             if ((fastafile != "") || (listfile != "") || (taxfile != "")) { 
329                 m->mothurOut("\n[NOTE]: The count file should contain only unique names, so mothur assumes your fasta, list and taxonomy files also contain only uniques.\n\n");
330             }
331         }
332                 
333                 //read through the correct file and output lines you want to keep
334                 if (namefile != "")                     {               readName();                     }
335                 if (fastafile != "")            {               readFasta();            }
336                 if (groupfile != "")            {               readGroup();            }
337         if (countfile != "")            {               readCount();            }
338                 if (alignfile != "")            {               readAlign();            }
339                 if (listfile != "")                     {               readList();                     }
340                 if (taxfile != "")                      {               readTax();                      }
341                 if (qualfile != "")                     {               readQual();                     }
342                 if (accnosfile2 != "")          {               compareAccnos();        }
343         
344         if (m->debug) { runSanityCheck(); }
345                 
346                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);  } return 0; }
347                 
348                 
349                 if (outputNames.size() != 0) {
350                         m->mothurOutEndLine();
351                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
352                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
353                         m->mothurOutEndLine();
354                         
355                         //set fasta file as new current fastafile
356                         string current = "";
357                         itTypes = outputTypes.find("fasta");
358                         if (itTypes != outputTypes.end()) {
359                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
360                         }
361                         
362                         itTypes = outputTypes.find("name");
363                         if (itTypes != outputTypes.end()) {
364                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
365                         }
366                         
367                         itTypes = outputTypes.find("group");
368                         if (itTypes != outputTypes.end()) {
369                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
370                         }
371                         
372                         itTypes = outputTypes.find("list");
373                         if (itTypes != outputTypes.end()) {
374                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
375                         }
376                         
377                         itTypes = outputTypes.find("taxonomy");
378                         if (itTypes != outputTypes.end()) {
379                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
380                         }
381                         
382                         itTypes = outputTypes.find("qfile");
383                         if (itTypes != outputTypes.end()) {
384                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
385                         }
386                         
387             itTypes = outputTypes.find("count");
388                         if (itTypes != outputTypes.end()) {
389                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
390                         }
391                 }
392                 
393                 return 0;               
394         }
395
396         catch(exception& e) {
397                 m->errorOut(e, "GetSeqsCommand", "execute");
398                 exit(1);
399         }
400 }
401
402 //**********************************************************************************************************************
403 int GetSeqsCommand::readFasta(){
404         try {
405                 string thisOutputDir = outputDir;
406                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
407                 map<string, string> variables; 
408         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(fastafile));
409         variables["[extension]"] = m->getExtension(fastafile);
410                 string outputFileName = getOutputFileName("fasta", variables);
411                 ofstream out;
412                 m->openOutputFile(outputFileName, out);
413                 
414                 
415                 ifstream in;
416                 m->openInputFile(fastafile, in);
417                 string name;
418                 
419                 bool wroteSomething = false;
420                 int selectedCount = 0;
421         
422         if (m->debug) { set<string> temp; sanity["fasta"] = temp; }
423                 
424                 while(!in.eof()){
425                 
426                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
427                         
428                         Sequence currSeq(in);
429                         name = currSeq.getName();
430             
431             if (!dups) {//adjust name if needed
432                 map<string, string>::iterator it = uniqueMap.find(name);
433                 if (it != uniqueMap.end()) { currSeq.setName(it->second); }
434             }
435                         
436             name = currSeq.getName();
437             
438                         if (name != "") {
439                                 //if this name is in the accnos file
440                                 if (names.count(name) != 0) {
441                                         wroteSomething = true;
442                                         
443                                         currSeq.printSequence(out);
444                                         selectedCount++;
445                     
446                     if (m->debug) { sanity["fasta"].insert(name); }
447                                 }
448                         }
449                         m->gobble(in);
450                 }
451                 in.close();     
452                 out.close();
453                 
454                 
455                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
456                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName); 
457                 
458                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
459                 
460                 return 0;
461
462         }
463         catch(exception& e) {
464                 m->errorOut(e, "GetSeqsCommand", "readFasta");
465                 exit(1);
466         }
467 }
468 //**********************************************************************************************************************
469 int GetSeqsCommand::readQual(){
470         try {
471                 string thisOutputDir = outputDir;
472                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
473                 map<string, string> variables; 
474         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(qualfile));
475         variables["[extension]"] = m->getExtension(qualfile);
476                 string outputFileName = getOutputFileName("qfile", variables);
477                 ofstream out;
478                 m->openOutputFile(outputFileName, out);
479                 
480                 
481                 ifstream in;
482                 m->openInputFile(qualfile, in);
483                 string name;
484                 
485                 bool wroteSomething = false;
486                 int selectedCount = 0;
487                 
488         if (m->debug) { set<string> temp; sanity["qual"] = temp; }
489                 
490                 while(!in.eof()){       
491                         string saveName = "";
492                         string name = "";
493                         string scores = "";
494                         
495                         in >> name;
496             
497             if (!dups) {//adjust name if needed
498                 map<string, string>::iterator it = uniqueMap.find(name);
499                 if (it != uniqueMap.end()) { name = it->second; }
500             }
501                                 
502                         if (name.length() != 0) { 
503                                 saveName = name.substr(1);
504                                 while (!in.eof())       {       
505                                         char c = in.get(); 
506                                         if (c == 10 || c == 13 || c == -1){     break;  }
507                                         else { name += c; }     
508                                 } 
509                                 m->gobble(in);
510                         }
511                         
512                         while(in){
513                                 char letter= in.get();
514                                 if(letter == '>'){      in.putback(letter);     break;  }
515                                 else{ scores += letter; }
516                         }
517                         
518                         m->gobble(in);
519                         
520                         if (names.count(saveName) != 0) {
521                                 wroteSomething = true;
522                                                 
523                                 out << name << endl << scores;
524                                 selectedCount++;
525                 if (m->debug) { sanity["qual"].insert(name); }
526                         }
527                         
528                         m->gobble(in);
529                 }
530                 in.close();
531                 out.close();
532                 
533                 
534                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
535                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName); 
536                 
537                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your quality file."); m->mothurOutEndLine();
538
539                 
540                 return 0;
541                 
542         }
543         catch(exception& e) {
544                 m->errorOut(e, "GetSeqsCommand", "readQual");
545                 exit(1);
546         }
547 }
548 //**********************************************************************************************************************
549 int GetSeqsCommand::readCount(){
550         try {
551                 string thisOutputDir = outputDir;
552                 if (outputDir == "") {  thisOutputDir += m->hasPath(countfile);  }
553                 map<string, string> variables; 
554                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
555         variables["[extension]"] = m->getExtension(countfile);
556                 string outputFileName = getOutputFileName("count", variables);
557                 
558                 ofstream out;
559                 m->openOutputFile(outputFileName, out);
560                 
561                 ifstream in;
562                 m->openInputFile(countfile, in);
563                 
564                 bool wroteSomething = false;
565                 int selectedCount = 0;
566                 
567         string headers = m->getline(in); m->gobble(in);
568         out << headers << endl;
569         
570         string name, rest; int thisTotal;
571         while (!in.eof()) {
572             
573             if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
574             
575             in >> name; m->gobble(in); 
576             in >> thisTotal; m->gobble(in);
577             rest = m->getline(in); m->gobble(in);
578             if (m->debug) { m->mothurOut("[DEBUG]: " + name + '\t' + rest + "\n"); }
579             
580             if (names.count(name) != 0) {
581                 out << name << '\t' << thisTotal << '\t' << rest << endl;
582                 wroteSomething = true;
583                 selectedCount+= thisTotal;
584             }
585         }
586         in.close();
587                 out.close();
588         
589         //check for groups that have been eliminated
590         CountTable ct;
591         if (ct.testGroups(outputFileName)) {
592             ct.readTable(outputFileName, true);
593             ct.printTable(outputFileName);
594         }
595                 
596                 if (wroteSomething == false) {  m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
597                 outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
598                 
599                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your count file."); m->mothurOutEndLine();
600         
601                 return 0;
602         }
603         catch(exception& e) {
604                 m->errorOut(e, "GetSeqsCommand", "readCount");
605                 exit(1);
606         }
607 }
608
609 //**********************************************************************************************************************
610 int GetSeqsCommand::readList(){
611         try {
612                 string thisOutputDir = outputDir;
613                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
614         map<string, string> variables; 
615                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
616         variables["[extension]"] = m->getExtension(listfile);
617                 string outputFileName = getOutputFileName("list", variables);
618                 ofstream out;
619                 m->openOutputFile(outputFileName, out);
620                 
621                 ifstream in;
622                 m->openInputFile(listfile, in);
623                 
624                 bool wroteSomething = false;
625                 int selectedCount = 0;
626         
627         if (m->debug) { set<string> temp; sanity["list"] = temp; }
628                 
629                 while(!in.eof()){
630                         
631                         selectedCount = 0;
632                         
633                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
634
635                         //read in list vector
636                         ListVector list(in);
637                         
638                         //make a new list vector
639                         ListVector newList;
640                         newList.setLabel(list.getLabel());
641                         
642                         //for each bin
643                         for (int i = 0; i < list.getNumBins(); i++) {
644                         
645                                 //parse out names that are in accnos file
646                                 string binnames = list.get(i);
647                 vector<string> bnames;
648                 m->splitAtComma(binnames, bnames);
649                                 
650                                 string newNames = "";
651                 for (int i = 0; i < bnames.size(); i++) {
652                                         string name = bnames[i];
653                                         //if that name is in the .accnos file, add it
654                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++; if (m->debug) { sanity["list"].insert(name); } }
655                                 }
656                         
657                                 //if there are names in this bin add to new list
658                                 if (newNames != "") { 
659                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
660                                         newList.push_back(newNames);    
661                                 }
662                         }
663                                 
664                         //print new listvector
665                         if (newList.getNumBins() != 0) {
666                                 wroteSomething = true;
667                                 newList.print(out);
668                         }
669                         
670                         m->gobble(in);
671                 }
672                 in.close();     
673                 out.close();
674                 
675                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
676                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
677                 
678                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
679                 
680                 return 0;
681
682         }
683         catch(exception& e) {
684                 m->errorOut(e, "GetSeqsCommand", "readList");
685                 exit(1);
686         }
687 }
688 //**********************************************************************************************************************
689 int GetSeqsCommand::readName(){
690         try {
691                 string thisOutputDir = outputDir;
692                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
693         map<string, string> variables; 
694                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(namefile));
695         variables["[extension]"] = m->getExtension(namefile);
696                 string outputFileName = getOutputFileName("name", variables);
697                 ofstream out;
698                 m->openOutputFile(outputFileName, out);
699                 
700
701                 ifstream in;
702                 m->openInputFile(namefile, in);
703                 string name, firstCol, secondCol;
704                 
705                 bool wroteSomething = false;
706                 int selectedCount = 0;
707         
708         if (m->debug) { set<string> temp; sanity["name"] = temp; }
709         if (m->debug) { set<string> temp; sanity["dupname"] = temp; }
710                 
711                 while(!in.eof()){
712                 
713                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
714
715                         in >> firstCol;                 m->gobble(in);
716                         in >> secondCol;
717                         
718                         string hold = "";
719                         if (dups) { hold = secondCol; }
720                         
721                         vector<string> parsedNames;
722                         m->splitAtComma(secondCol, parsedNames);
723                         
724                         vector<string> validSecond;
725                         for (int i = 0; i < parsedNames.size(); i++) {
726                                 if (names.count(parsedNames[i]) != 0) {
727                                         validSecond.push_back(parsedNames[i]);
728                     if (m->debug) { sanity["dupname"].insert(parsedNames[i]); }
729                                 }
730                         }
731
732                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
733                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]); if (m->debug) { sanity["dupname"].insert(parsedNames[i]); } }
734                                 out << firstCol << '\t' << hold << endl;
735                                 wroteSomething = true;
736                                 selectedCount += parsedNames.size();
737                 if (m->debug) { sanity["name"].insert(firstCol); }
738                         }else {
739                 
740                                 selectedCount += validSecond.size();
741                                 
742                                 //if the name in the first column is in the set then print it and any other names in second column also in set
743                                 if (names.count(firstCol) != 0) {
744                                 
745                                         wroteSomething = true;
746                                         
747                                         out << firstCol << '\t';
748                                         
749                                         //you know you have at least one valid second since first column is valid
750                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
751                                         out << validSecond[validSecond.size()-1] << endl;
752                     
753                     if (m->debug) { sanity["name"].insert(firstCol); }
754                                         
755                                 
756                                 //make first name in set you come to first column and then add the remaining names to second column
757                                 }else {
758                     
759                                         //you want part of this row
760                                         if (validSecond.size() != 0) {
761                                         
762                                                 wroteSomething = true;
763                                                 
764                                                 out << validSecond[0] << '\t';
765                         //we are changing the unique name in the fasta file
766                         uniqueMap[firstCol] = validSecond[0];
767                                         
768                                                 //you know you have at least one valid second since first column is valid
769                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
770                                                 out << validSecond[validSecond.size()-1] << endl;
771                         
772                         if (m->debug) { sanity["name"].insert(validSecond[0]); }
773                                         }
774                                 }
775                         }
776                         m->gobble(in);
777                 }
778                 in.close();
779                 out.close();
780                 
781                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
782                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
783                 
784                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
785                 
786                 return 0;
787                 
788         }
789         catch(exception& e) {
790                 m->errorOut(e, "GetSeqsCommand", "readName");
791                 exit(1);
792         }
793 }
794
795 //**********************************************************************************************************************
796 int GetSeqsCommand::readGroup(){
797         try {
798                 string thisOutputDir = outputDir;
799                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
800                 map<string, string> variables; 
801                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(groupfile));
802         variables["[extension]"] = m->getExtension(groupfile);
803                 string outputFileName = getOutputFileName("group", variables);
804                 ofstream out;
805                 m->openOutputFile(outputFileName, out);
806                 
807
808                 ifstream in;
809                 m->openInputFile(groupfile, in);
810                 string name, group;
811                 
812                 bool wroteSomething = false;
813                 int selectedCount = 0;
814         
815         if (m->debug) { set<string> temp; sanity["group"] = temp; }
816                 
817                 while(!in.eof()){
818
819                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
820
821
822                         in >> name;                             //read from first column
823                         in >> group;                    //read from second column
824             
825                         
826                         //if this name is in the accnos file
827                         if (names.count(name) != 0) {
828                                 wroteSomething = true;
829                                 
830                                 out << name << '\t' << group << endl;
831                                 selectedCount++;
832                 
833                 if (m->debug) {  sanity["group"].insert(name); }
834                         }
835                                         
836                         m->gobble(in);
837                 }
838                 in.close();
839                 out.close();
840                 
841                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
842                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
843                 
844                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
845
846                 
847                 return 0;
848
849         }
850         catch(exception& e) {
851                 m->errorOut(e, "GetSeqsCommand", "readGroup");
852                 exit(1);
853         }
854 }
855 //**********************************************************************************************************************
856 int GetSeqsCommand::readTax(){
857         try {
858                 string thisOutputDir = outputDir;
859                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
860                 map<string, string> variables; 
861                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(taxfile));
862         variables["[extension]"] = m->getExtension(taxfile);
863                 string outputFileName = getOutputFileName("taxonomy", variables);
864                 ofstream out;
865                 m->openOutputFile(outputFileName, out);
866                 
867                 ifstream in;
868                 m->openInputFile(taxfile, in);
869                 string name, tax;
870                 
871                 bool wroteSomething = false;
872                 int selectedCount = 0;
873         
874         if (m->debug) { set<string> temp; sanity["tax"] = temp; }
875                 
876                 while(!in.eof()){
877
878                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
879
880                         in >> name;                             //read from first column
881                         in >> tax;                      //read from second column
882             
883             if (!dups) {//adjust name if needed
884                 map<string, string>::iterator it = uniqueMap.find(name);
885                 if (it != uniqueMap.end()) { name = it->second; }
886             }
887                         
888                         //if this name is in the accnos file
889                         if (names.count(name) != 0) {
890                                 wroteSomething = true;
891                                 
892                                 out << name << '\t' << tax << endl;
893                                 selectedCount++;
894                 
895                 if (m->debug) { sanity["tax"].insert(name); }
896                         }
897                                         
898                         m->gobble(in);
899                 }
900                 in.close();
901                 out.close();
902                 
903                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
904                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
905                 
906                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
907                         
908                 return 0;
909
910         }
911         catch(exception& e) {
912                 m->errorOut(e, "GetSeqsCommand", "readTax");
913                 exit(1);
914         }
915 }
916 //**********************************************************************************************************************
917 //alignreport file has a column header line then all other lines contain 16 columns.  we just want the first column since that contains the name
918 int GetSeqsCommand::readAlign(){
919         try {
920                 string thisOutputDir = outputDir;
921                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
922         map<string, string> variables; 
923                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(alignfile));
924                 string outputFileName = getOutputFileName("alignreport", variables);
925                 ofstream out;
926                 m->openOutputFile(outputFileName, out);
927                 
928
929                 ifstream in;
930                 m->openInputFile(alignfile, in);
931                 string name, junk;
932                 
933                 bool wroteSomething = false;
934                 int selectedCount = 0;
935                 
936                 //read column headers
937                 for (int i = 0; i < 16; i++) {  
938                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
939                         else                    {       break;                  }
940                 }
941                 out << endl;
942                 
943                 while(!in.eof()){
944                 
945                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
946
947
948                         in >> name;                             //read from first column
949             
950             if (!dups) {//adjust name if needed
951                 map<string, string>::iterator it = uniqueMap.find(name);
952                 if (it != uniqueMap.end()) { name = it->second; }
953             }
954                         
955                         //if this name is in the accnos file
956                         if (names.count(name) != 0) {
957                                 wroteSomething = true;
958                                 selectedCount++;
959                                 
960                                 out << name << '\t';
961                                 
962                                 //read rest
963                                 for (int i = 0; i < 15; i++) {  
964                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
965                                         else                    {       break;                  }
966                                 }
967                                 out << endl;
968                                 
969                         }else {//still read just don't do anything with it
970                                 //read rest
971                                 for (int i = 0; i < 15; i++) {  
972                                         if (!in.eof())  {       in >> junk;             }
973                                         else                    {       break;                  }
974                                 }
975                         }
976                         
977                         m->gobble(in);
978                 }
979                 in.close();
980                 out.close();
981                 
982                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
983                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
984                 
985                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
986                 
987                 return 0;
988                 
989         }
990         catch(exception& e) {
991                 m->errorOut(e, "GetSeqsCommand", "readAlign");
992                 exit(1);
993         }
994 }
995 //**********************************************************************************************************************
996 //just looking at common mistakes. 
997 int GetSeqsCommand::runSanityCheck(){
998         try {
999         string thisOutputDir = outputDir;
1000         if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
1001         string filename = outputDir + "get.seqs.debug.report";
1002         
1003         ofstream out;
1004                 m->openOutputFile(filename, out); 
1005
1006        
1007         //compare fasta, name, qual and taxonomy if given to make sure they contain the same seqs
1008         if (fastafile != "") {
1009             if (namefile != "") { //compare with fasta
1010                 if (sanity["fasta"] != sanity["name"]) { //create mismatch file
1011                     createMisMatchFile(out, fastafile, namefile, sanity["fasta"], sanity["name"]);
1012                 }
1013             }
1014             if (qualfile != "") {
1015                 if (sanity["fasta"] != sanity["qual"]) { //create mismatch file
1016                     createMisMatchFile(out, fastafile, qualfile, sanity["fasta"], sanity["qual"]);
1017                 }
1018             }
1019             if (taxfile != "") {
1020                 if (sanity["fasta"] != sanity["tax"]) { //create mismatch file
1021                     createMisMatchFile(out, fastafile, taxfile, sanity["fasta"], sanity["tax"]);
1022                 }
1023             }
1024         }
1025         
1026         //compare dupnames, groups and list if given to make sure they match
1027         if (namefile != "") {
1028             if (groupfile != "") {
1029                 if (sanity["dupname"] != sanity["group"]) { //create mismatch file
1030                     createMisMatchFile(out, namefile, groupfile, sanity["dupname"], sanity["group"]);
1031                 } 
1032             }
1033             if (listfile != "") {
1034                 if (sanity["dupname"] != sanity["list"]) { //create mismatch file
1035                     createMisMatchFile(out, namefile, listfile, sanity["dupname"], sanity["list"]);
1036                 } 
1037             }
1038         }else{
1039
1040             if ((groupfile != "") && (fastafile != "")) {
1041                 if (sanity["fasta"] != sanity["group"]) { //create mismatch file
1042                     createMisMatchFile(out, fastafile, groupfile, sanity["fasta"], sanity["group"]);
1043                 } 
1044             }
1045         }
1046         
1047         out.close();
1048         
1049         if (m->isBlank(filename)) { m->mothurRemove(filename); }
1050         else { m->mothurOut("\n[DEBUG]: " + filename + " contains the file mismatches.\n");outputNames.push_back(filename); outputTypes["debug"].push_back(filename); }
1051         
1052         return 0;
1053     }
1054         catch(exception& e) {
1055                 m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
1056                 exit(1);
1057         }
1058 }
1059 //**********************************************************************************************************************
1060 //just looking at common mistakes. 
1061 int GetSeqsCommand::createMisMatchFile(ofstream& out, string filename1, string filename2, set<string> set1, set<string> set2){
1062         try {
1063         out << "****************************************" << endl << endl;
1064         out << "Names unique to " << filename1 << ":\n";
1065         
1066         //remove names in set1 that are also in set2
1067         for (set<string>::iterator it = set1.begin(); it != set1.end();) {
1068             string name = *it;
1069             
1070             if (set2.count(name) == 0)  { out << name << endl;  } //name unique to set1
1071             else                        { set2.erase(name);     } //you are in both so erase 
1072             set1.erase(it++);
1073         }
1074         
1075         out << "\nNames unique to " << filename2 << ":\n";
1076         //output results
1077         for (set<string>::iterator it = set2.begin(); it != set2.end(); it++) {  out << *it << endl;  }
1078         
1079         out << "****************************************" << endl << endl;
1080         
1081         return 0;
1082     }
1083         catch(exception& e) {
1084                 m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
1085                 exit(1);
1086         }
1087 }
1088 //**********************************************************************************************************************
1089
1090 int GetSeqsCommand::compareAccnos(){
1091         try {
1092                 
1093                 string thisOutputDir = outputDir;
1094                 if (outputDir == "") {  thisOutputDir += m->hasPath(accnosfile);  }
1095         map<string, string> variables; 
1096                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile));
1097                 string outputFileName = getOutputFileName("accnosreport", variables);
1098                 
1099                 ofstream out;
1100                 m->openOutputFile(outputFileName, out);
1101                 
1102                 ifstream in;
1103                 m->openInputFile(accnosfile2, in);
1104                 string name;
1105                 
1106                 set<string> namesAccnos2;
1107                 set<string> namesDups;
1108                 set<string> namesAccnos = names;
1109                 
1110                 map<string, int> nameCount;
1111                 
1112                 if (namefile != "") {
1113                         ifstream inName;
1114                         m->openInputFile(namefile, inName);
1115                         
1116                         
1117                         while(!inName.eof()){
1118                                 
1119                                 if (m->control_pressed) { inName.close(); return 0; }
1120                                 
1121                                 string thisname, repnames;
1122                                 
1123                                 inName >> thisname;             m->gobble(inName);              //read from first column
1124                                 inName >> repnames;                     //read from second column
1125                                 
1126                                 int num = m->getNumNames(repnames);
1127                                 nameCount[thisname] = num;
1128                                 
1129                                 m->gobble(inName);
1130                         }
1131                         inName.close(); 
1132                 }
1133                 
1134                 while(!in.eof()){
1135                         in >> name;
1136                         
1137                         if (namesAccnos.count(name) == 0){ //name unique to accnos2
1138                                 int pos = name.find_last_of('_');
1139                                 string tempName = name;
1140                                 if (pos != string::npos) {  tempName = tempName.substr(pos+1); cout << tempName << endl; }
1141                                 if (namesAccnos.count(tempName) == 0){
1142                                         namesAccnos2.insert(name);
1143                                 }else { //you are in both so erase
1144                                         namesAccnos.erase(name);
1145                                         namesDups.insert(name);
1146                                 }
1147                         }else { //you are in both so erase
1148                                 namesAccnos.erase(name);
1149                                 namesDups.insert(name);
1150                         }
1151                         
1152                         m->gobble(in);
1153                 }
1154                 in.close();     
1155                 
1156                 out << "Names in both files : " + toString(namesDups.size()) << endl;
1157                 m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
1158                 
1159                 for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
1160                         out << (*it);
1161                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1162                         out << endl;
1163                 }
1164                 
1165                 out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
1166                 m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
1167                 
1168                 for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
1169                         out << (*it);
1170                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1171                         out << endl;
1172                 }
1173                 
1174                 out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
1175                 m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
1176                 
1177                 for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
1178                         out << (*it);
1179                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1180                         out << endl;
1181                 }
1182
1183                 out.close(); 
1184                 
1185                 outputNames.push_back(outputFileName);  outputTypes["accnosreport"].push_back(outputFileName);
1186                 
1187                 return 0;
1188                 
1189         }
1190         catch(exception& e) {
1191                 m->errorOut(e, "GetSeqsCommand", "compareAccnos");
1192                 exit(1);
1193         }
1194 }
1195
1196
1197 //**********************************************************************************************************************
1198