]> git.donarmstrong.com Git - mothur.git/blob - pairwiseseqscommand.h
Merge remote-tracking branch 'origin/master'
[mothur.git] / pairwiseseqscommand.h
1 #ifndef PAIRWISESEQSCOMMAND_H
2 #define PAIRWISESEQSCOMMAND_H
3
4 /*
5  *  pairwiseseqscommand.h
6  *  Mothur
7  *
8  *  Created by westcott on 10/20/10.
9  *  Copyright 2010 Schloss Lab. All rights reserved.
10  *
11  */
12
13 #include "mothur.h"
14 #include "command.hpp"
15 #include "database.hpp"
16 #include "alignment.hpp"
17 #include "validcalculator.h"
18 #include "dist.h"
19 #include "sequencedb.h"
20 #include "sequence.hpp"
21
22 #include "gotohoverlap.hpp"
23 #include "needlemanoverlap.hpp"
24 #include "blastalign.hpp"
25 #include "noalign.hpp"
26
27 #include "ignoregaps.h"
28 #include "eachgapdist.h"
29 #include "eachgapignore.h"
30 #include "onegapdist.h"
31 #include "onegapignore.h"
32
33 class PairwiseSeqsCommand : public Command {
34         
35 public:
36         PairwiseSeqsCommand(string);    
37         PairwiseSeqsCommand();
38         ~PairwiseSeqsCommand() {}
39         
40         vector<string> setParameters();
41         string getCommandName()                 { return "pairwise.seqs";               }
42         string getCommandCategory()             { return "Sequence Processing"; }
43         
44         string getHelpString(); 
45     string getOutputPattern(string);    
46         string getCitation() { return "Needleman SB, Wunsch CD (1970). A general method applicable to the search for similarities in the amino acid sequence of two proteins. J Mol Biol 48: 443-53. [ for needleman ]\nGotoh O (1982). An improved algorithm for matching biological sequences. J Mol Biol 162: 705-8. [ for gotoh ] \nhttp://www.mothur.org/wiki/Pairwise.seqs"; }
47         string getDescription()         { return "calculates pairwise distances from an unaligned fasta file"; }
48
49         int execute(); 
50         void help() { m->mothurOut(getHelpString()); }  
51         
52 private:
53         struct distlinePair {
54                 int start;
55                 int end;
56         };
57         
58         vector<int> processIDS;   //end line, processid
59         vector<distlinePair> lines;
60         
61         SequenceDB alignDB;
62         
63         void createProcesses(string);
64         int driver(int, int, string, float);
65         int driver(int, int, string, string);
66         
67         #ifdef USE_MPI 
68         int driverMPI(int, int, MPI_File&, float);
69         int driverMPI(int, int, string, unsigned long long&);
70         int driverMPI(int, int, string, unsigned long long&, string);
71         #endif
72         
73         string fastaFileName, align, calc, outputDir, output;
74         float match, misMatch, gapOpen, gapExtend, cutoff;
75         int processors, longestBase;
76         vector<string> fastaFileNames, Estimators;
77         vector<string> outputNames;
78         
79         bool abort, countends, compress;
80 };
81
82 /**************************************************************************************************/
83 //custom data structure for threads to use.
84 // This is passed by void pointer so it can be any data type
85 // that can be passed using a single void pointer (LPVOID).
86 struct pairwiseData {
87     string outputFileName;
88         string align, square, distcalcType, output;
89         unsigned long long start;
90         unsigned long long end;
91         MothurOut* m;
92         float match, misMatch, gapOpen, gapExtend, cutoff;
93         int count, threadID, longestBase;
94     bool countends;
95     SequenceDB alignDB;
96         
97         pairwiseData(){}
98         pairwiseData(string ofn, string al, string sq, string di, bool co, string op, SequenceDB DB, MothurOut* mout, unsigned long long st, unsigned long long en, float ma, float misMa, float gapO, float gapE, int thr, int tid) {
99                 outputFileName = ofn;
100                 m = mout;
101                 start = st;
102                 end = en;
103                 match = ma; 
104                 misMatch = misMa;
105                 gapOpen = gapO; 
106                 gapExtend = gapE; 
107                 longestBase = thr;
108                 align = al;
109         square = sq;
110         distcalcType = di;
111         countends = co;
112         alignDB = DB;
113                 count = 0;
114         output = op;
115                 threadID = tid;
116         }
117 };
118
119 /**************************************************************************************************/
120 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
121 #else
122 static DWORD WINAPI MyPairwiseSquareThreadFunction(LPVOID lpParam){ 
123         pairwiseData* pDataArray;
124         pDataArray = (pairwiseData*)lpParam;
125         
126         try {
127                 ofstream outFile((pDataArray->outputFileName).c_str(), ios::trunc);
128                 outFile.setf(ios::fixed, ios::showpoint);
129                 outFile << setprecision(4);
130                 
131                 pDataArray->count = pDataArray->end;
132         
133         int startTime = time(NULL);
134         
135         Alignment* alignment;
136         if(pDataArray->align == "gotoh")                        {       alignment = new GotohOverlap(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch, pDataArray->longestBase);                     }
137                 else if(pDataArray->align == "needleman")       {       alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, pDataArray->longestBase);                                }
138                 else if(pDataArray->align == "blast")           {       alignment = new BlastAlignment(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch);            }
139                 else if(pDataArray->align == "noalign")         {       alignment = new NoAlign();                                                                                                      }
140                 else {
141                         pDataArray->m->mothurOut(pDataArray->align + " is not a valid alignment option. I will run the command using needleman.");
142                         pDataArray->m->mothurOutEndLine();
143                         alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, pDataArray->longestBase);
144                 }
145                 
146         ValidCalculators validCalculator;
147         Dist* distCalculator;
148         if (pDataArray->countends) {
149             if (validCalculator.isValidCalculator("distance", pDataArray->distcalcType) == true) { 
150                 if (pDataArray->distcalcType == "nogaps")                       {       distCalculator = new ignoreGaps();      }
151                 else if (pDataArray->distcalcType == "eachgap") {       distCalculator = new eachGapDist();     }
152                 else if (pDataArray->distcalcType == "onegap")          {       distCalculator = new oneGapDist();      }
153             }
154         }else {
155             if (validCalculator.isValidCalculator("distance", pDataArray->distcalcType) == true) { 
156                 if (pDataArray->distcalcType == "nogaps")               {       distCalculator = new ignoreGaps();                                      }
157                 else if (pDataArray->distcalcType == "eachgap"){        distCalculator = new eachGapIgnoreTermGapDist();        }
158                 else if (pDataArray->distcalcType == "onegap")  {       distCalculator = new oneGapIgnoreTermGapDist();         }
159             }
160         }
161
162         if(pDataArray->start == 0){     outFile << pDataArray->alignDB.getNumSeqs() << endl;    }
163                 
164                 for(int i=pDataArray->start;i<pDataArray->end;i++){
165             
166                         string name = pDataArray->alignDB.get(i).getName();
167                         //pad with spaces to make compatible
168                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
169             
170                         outFile << name << '\t';        
171                         
172                         for(int j=0;j<pDataArray->alignDB.getNumSeqs();j++){
173                                 
174                                 if (pDataArray->m->control_pressed) { outFile.close(); delete alignment; delete distCalculator; return 0;  }
175                                 
176                                 if (pDataArray->alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
177                                         alignment->resize(pDataArray->alignDB.get(i).getUnaligned().length()+1);
178                                 }
179                                 
180                                 if (pDataArray->alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
181                                         alignment->resize(pDataArray->alignDB.get(j).getUnaligned().length()+1);
182                                 }
183                                 
184                                 Sequence seqI(pDataArray->alignDB.get(i).getName(), pDataArray->alignDB.get(i).getAligned());
185                                 Sequence seqJ(pDataArray->alignDB.get(j).getName(), pDataArray->alignDB.get(j).getAligned());
186                                 
187                                 alignment->align(seqI.getUnaligned(), seqJ.getUnaligned());
188                                 seqI.setAligned(alignment->getSeqAAln());
189                                 seqJ.setAligned(alignment->getSeqBAln());
190                                 
191                                 distCalculator->calcDist(seqI, seqJ);
192                                 double dist = distCalculator->getDist();
193                 
194                                 outFile << dist << '\t'; 
195                         }
196                         
197                         outFile << endl; 
198                         
199                         if(i % 100 == 0){
200                                 pDataArray->m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); pDataArray->m->mothurOutEndLine();
201                         }
202                         
203                 }
204                 pDataArray->m->mothurOut(toString(pDataArray->end-1) + "\t" + toString(time(NULL) - startTime)); pDataArray->m->mothurOutEndLine();
205                 
206                 outFile.close();
207         delete alignment;
208         delete distCalculator;
209
210         
211     }
212         catch(exception& e) {
213                 pDataArray->m->errorOut(e, "PairwiseSeqsCommand", "MyPairwiseSquareThreadFunction");
214                 exit(1);
215         }
216
217
218 /**************************************************************************************************/
219 static DWORD WINAPI MyPairwiseThreadFunction(LPVOID lpParam){ 
220         pairwiseData* pDataArray;
221         pDataArray = (pairwiseData*)lpParam;
222         
223         try {
224                 ofstream outFile((pDataArray->outputFileName).c_str(), ios::trunc);
225                 outFile.setf(ios::fixed, ios::showpoint);
226                 outFile << setprecision(4);
227                 
228         pDataArray->count = pDataArray->end;
229         
230         int startTime = time(NULL);
231         
232         Alignment* alignment;
233         if(pDataArray->align == "gotoh")                        {       alignment = new GotohOverlap(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch, pDataArray->longestBase);                     }
234                 else if(pDataArray->align == "needleman")       {       alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, pDataArray->longestBase);                                }
235                 else if(pDataArray->align == "blast")           {       alignment = new BlastAlignment(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch);            }
236                 else if(pDataArray->align == "noalign")         {       alignment = new NoAlign();                                                                                                      }
237                 else {
238                         pDataArray->m->mothurOut(pDataArray->align + " is not a valid alignment option. I will run the command using needleman.");
239                         pDataArray->m->mothurOutEndLine();
240                         alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, pDataArray->longestBase);
241                 }
242                 
243         ValidCalculators validCalculator;
244         Dist* distCalculator;
245         if (pDataArray->countends) {
246             if (validCalculator.isValidCalculator("distance", pDataArray->distcalcType) == true) { 
247                 if (pDataArray->distcalcType == "nogaps")                       {       distCalculator = new ignoreGaps();      }
248                 else if (pDataArray->distcalcType == "eachgap") {       distCalculator = new eachGapDist();     }
249                 else if (pDataArray->distcalcType == "onegap")          {       distCalculator = new oneGapDist();      }
250             }
251         }else {
252             if (validCalculator.isValidCalculator("distance", pDataArray->distcalcType) == true) { 
253                 if (pDataArray->distcalcType == "nogaps")               {       distCalculator = new ignoreGaps();                                      }
254                 else if (pDataArray->distcalcType == "eachgap"){        distCalculator = new eachGapIgnoreTermGapDist();        }
255                 else if (pDataArray->distcalcType == "onegap")  {       distCalculator = new oneGapIgnoreTermGapDist();         }
256             }
257         }
258         
259         if((pDataArray->output == "lt") && pDataArray->start == 0){     outFile << pDataArray->alignDB.getNumSeqs() << endl;    }
260                 
261                 for(int i=pDataArray->start;i<pDataArray->end;i++){
262             
263                         if(pDataArray->output == "lt")  {       
264                                 string name = pDataArray->alignDB.get(i).getName();
265                                 if (name.length() < 10) { //pad with spaces to make compatible
266                                         while (name.length() < 10) {  name += " ";  }
267                                 }
268                                 outFile << name << '\t';        
269                         }
270
271                         
272                         for(int j=0;j<i;j++){
273                                 
274                                 if (pDataArray->m->control_pressed) { outFile.close(); delete alignment; delete distCalculator; return 0;  }
275                                 
276                                 if (pDataArray->alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
277                                         alignment->resize(pDataArray->alignDB.get(i).getUnaligned().length()+1);
278                                 }
279                                 
280                                 if (pDataArray->alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
281                                         alignment->resize(pDataArray->alignDB.get(j).getUnaligned().length()+1);
282                                 }
283                                 
284                                 Sequence seqI(pDataArray->alignDB.get(i).getName(), pDataArray->alignDB.get(i).getAligned());
285                                 Sequence seqJ(pDataArray->alignDB.get(j).getName(), pDataArray->alignDB.get(j).getAligned());
286                                 
287                                 alignment->align(seqI.getUnaligned(), seqJ.getUnaligned());
288                                 seqI.setAligned(alignment->getSeqAAln());
289                                 seqJ.setAligned(alignment->getSeqBAln());
290                                 
291                                 distCalculator->calcDist(seqI, seqJ);
292                                 double dist = distCalculator->getDist();
293                 
294                                 if(dist <= pDataArray->cutoff){
295                                         if (pDataArray->output == "column") { outFile << pDataArray->alignDB.get(i).getName() << ' ' << pDataArray->alignDB.get(j).getName() << ' ' << dist << endl; }
296                                 }
297                                 if (pDataArray->output == "lt") {  outFile << dist << '\t'; }
298                         }
299                         
300                         if (pDataArray->output == "lt") { outFile << endl; }
301                         
302                         if(i % 100 == 0){
303                                 pDataArray->m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); pDataArray->m->mothurOutEndLine();
304                         }
305                         
306                 }
307                 pDataArray->m->mothurOut(toString(pDataArray->end-1) + "\t" + toString(time(NULL) - startTime)); pDataArray->m->mothurOutEndLine();
308                 
309                 outFile.close();
310         delete alignment;
311         delete distCalculator;
312         
313         
314     }
315         catch(exception& e) {
316                 pDataArray->m->errorOut(e, "PairwiseSeqsCommand", "MyPairwiseThreadFunction");
317                 exit(1);
318         }
319
320
321 #endif
322
323
324 #endif
325