5 * Created by Sarah Westcott on 5/7/09.
6 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
10 #include "distancecommand.h"
11 #include "ignoregaps.h"
12 #include "eachgapdist.h"
13 #include "eachgapignore.h"
14 #include "onegapdist.h"
15 #include "onegapignore.h"
17 //**********************************************************************************************************************
19 DistanceCommand::DistanceCommand(string option) {
24 //allow user to run help
25 if(option == "help") { help(); abort = true; }
28 //valid paramters for this command
29 string Array[] = {"fasta", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir"};
30 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
32 OptionParser parser(option);
33 map<string, string> parameters = parser.getParameters();
35 ValidParameters validParameter;
36 map<string, string>::iterator it2;
38 //check to make sure all parameters are valid for command
39 for (it2 = parameters.begin(); it2 != parameters.end(); it2++) {
40 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) { abort = true; }
43 //if the user changes the input directory command factory will send this info to us in the output parameter
44 string inputDir = validParameter.validFile(parameters, "inputdir", false);
45 if (inputDir == "not found"){ inputDir = ""; }
48 it2 = parameters.find("fasta");
49 //user has given a template file
50 if(it2 != parameters.end()){
51 path = hasPath(it2->second);
52 //if the user has not given a path then, add inputdir. else leave path alone.
53 if (path == "") { parameters["fasta"] = inputDir + it2->second; }
57 //check for required parameters
58 fastafile = validParameter.validFile(parameters, "fasta", true);
59 if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the dist.seqs command."); m->mothurOutEndLine(); abort = true; }
60 else if (fastafile == "not open") { abort = true; }
63 openInputFile(fastafile, inFASTA);
64 alignDB = SequenceDB(inFASTA);
68 //if the user changes the output directory command factory will send this info to us in the output parameter
69 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
71 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
74 //check for optional parameter and set defaults
75 // ...at some point should added some additional type checking...
76 calc = validParameter.validFile(parameters, "calc", false);
77 if (calc == "not found") { calc = "onegap"; }
79 if (calc == "default") { calc = "onegap"; }
81 splitAtDash(calc, Estimators);
84 temp = validParameter.validFile(parameters, "countends", false); if(temp == "not found"){ temp = "T"; }
85 convert(temp, countends);
87 temp = validParameter.validFile(parameters, "cutoff", false); if(temp == "not found"){ temp = "1.0"; }
88 convert(temp, cutoff);
90 temp = validParameter.validFile(parameters, "processors", false); if(temp == "not found"){ temp = "1"; }
91 convert(temp, processors);
93 output = validParameter.validFile(parameters, "output", false); if(output == "not found"){ output = "column"; }
95 if ((output != "column") && (output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are column, lt and square. I will use column."); m->mothurOutEndLine(); output = "column"; }
97 ValidCalculators validCalculator;
99 if (isTrue(countends) == true) {
100 for (int i=0; i<Estimators.size(); i++) {
101 if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
102 if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
103 else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
104 else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
108 for (int i=0; i<Estimators.size(); i++) {
109 if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
110 if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
111 else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
112 else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
120 catch(exception& e) {
121 m->errorOut(e, "DistanceCommand", "DistanceCommand");
126 //**********************************************************************************************************************
128 DistanceCommand::~DistanceCommand(){
130 for(int i=0;i<lines.size();i++){
136 //**********************************************************************************************************************
138 void DistanceCommand::help(){
140 m->mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
141 m->mothurOut("The dist.seqs command parameters are fasta, calc, countends, output, cutoff and processors. \n");
142 m->mothurOut("The fasta parameter is required.\n");
143 m->mothurOut("The calc parameter allows you to specify the method of calculating the distances. Your options are: nogaps, onegap or eachgap. The default is onegap.\n");
144 m->mothurOut("The countends parameter allows you to specify whether to include terminal gaps in distance. Your options are: T or F. The default is T.\n");
145 m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
146 m->mothurOut("The output parameter allows you to specify format of your distance matrix. Options are column, lt, and square. The default is column.\n");
147 m->mothurOut("The processors parameter allows you to specify number of processors to use. The default is 1.\n");
148 m->mothurOut("The dist.seqs command should be in the following format: \n");
149 m->mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
150 m->mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
151 m->mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
153 catch(exception& e) {
154 m->errorOut(e, "DistanceCommand", "help");
158 //**********************************************************************************************************************
160 int DistanceCommand::execute(){
163 if (abort == true) { return 0; }
165 int startTime = time(NULL);
167 int numSeqs = alignDB.getNumSeqs();
172 if (output == "lt") { //does the user want lower triangle phylip formatted file
173 outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "phylip.dist";
174 remove(outputFile.c_str());
176 //output numSeqs to phylip formatted dist file
177 }else if (output == "column") { //user wants column format
178 outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "dist";
179 remove(outputFile.c_str());
180 }else { //assume square
181 outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "square.dist";
182 remove(outputFile.c_str());
192 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
193 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
195 //each process gets where it should start and stop in the file
196 start = int (sqrt(float(pid)/float(processors)) * numSeqs);
197 end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
199 if (output != "lt") {
201 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
203 char* filename = new char[outputFile.length()];
\r
204 memcpy(filename, outputFile.c_str(), outputFile.length());
206 MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
209 if (pid == 0) { //you are the root process
213 driverMPI(start, end, outMPI, cutoff);
215 if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
218 for(int i = 1; i < processors; i++) {
219 if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
222 MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
224 }else { //you are a child process
226 driverMPI(start, end, outMPI, cutoff);
228 if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
232 //tell parent you are done.
233 MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
236 MPI_File_close(&outMPI);
238 }else { //lower triangle format
239 if (pid == 0) { //you are the root process
244 driverMPI(start, end, outputFile, mySize);
246 if (m->control_pressed) { delete distCalculator; return 0; }
248 int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
252 char* filename = new char[outputFile.length()];
\r
253 memcpy(filename, outputFile.c_str(), outputFile.length());
255 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
259 for(int b = 1; b < processors; b++) {
262 if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
264 MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
266 string outTemp = outputFile + toString(b) + ".temp";
268 char* buf = new char[outTemp.length()];
\r
269 memcpy(buf, outTemp.c_str(), outTemp.length());
271 MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
275 while (count < fileSize) { //read 1000 characters at a time
278 MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
279 MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
283 MPI_File_close(&inMPI); //deleted on close
286 MPI_File_close(&outMPI);
287 }else { //you are a child process
290 driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size);
292 if (m->control_pressed) { delete distCalculator; return 0; }
294 //tell parent you are done.
295 MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
300 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
301 //if you don't need to fork anything
303 driver(0, numSeqs, outputFile, cutoff);
304 }else{ //you have multiple processors
306 for (int i = 0; i < processors; i++) {
307 lines.push_back(new linePair());
308 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
309 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
312 createProcesses(outputFile);
314 map<int, int>::iterator it = processIDS.begin();
315 rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
318 //append and remove temp files
319 for (; it != processIDS.end(); it++) {
320 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
321 remove((outputFile + toString(it->second) + ".temp").c_str());
326 driver(0, numSeqs, outputFile, cutoff);
330 if (m->control_pressed) { delete distCalculator; remove(outputFile.c_str()); return 0; }
333 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
335 if (pid == 0) { //only one process should output to screen
338 if (output == "square") { convertMatrix(outputFile); }
344 if (m->control_pressed) { delete distCalculator; remove(outputFile.c_str()); return 0; }
346 delete distCalculator;
348 m->mothurOutEndLine();
349 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
350 m->mothurOut(outputFile); m->mothurOutEndLine();
351 m->mothurOutEndLine();
352 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
356 catch(exception& e) {
357 m->errorOut(e, "DistanceCommand", "execute");
361 /**************************************************************************************************/
362 void DistanceCommand::createProcesses(string filename) {
364 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
368 //loop through and create all the processes you want
369 while (process != processors) {
373 processIDS[lines[process]->end] = pid; //create map from line number to pid so you can append files in correct order later
376 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
378 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
381 //force parent to wait until all the processes are done
382 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
383 int temp = it->second;
388 catch(exception& e) {
389 m->errorOut(e, "DistanceCommand", "createProcesses");
394 /**************************************************************************************************/
395 /////// need to fix to work with calcs and sequencedb
396 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
399 int startTime = time(NULL);
402 ofstream outFile(dFileName.c_str(), ios::trunc);
403 outFile.setf(ios::fixed, ios::showpoint);
404 outFile << setprecision(4);
406 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl; }
408 for(int i=startLine;i<endLine;i++){
410 string name = alignDB.get(i).getName();
411 if (name.length() < 10) { //pad with spaces to make compatible
412 while (name.length() < 10) { name += " "; }
414 outFile << name << '\t';
416 for(int j=0;j<i;j++){
418 if (m->control_pressed) { outFile.close(); return 0; }
420 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
421 double dist = distCalculator->getDist();
424 if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
426 if (output == "lt") { outFile << dist << '\t'; }
428 if (output == "square") { //make a square column you can convert to square phylip
429 outFile << alignDB.get(i).getName() << '\t' << alignDB.get(j).getName() << '\t' << dist << endl;
430 outFile << alignDB.get(j).getName() << '\t' << alignDB.get(i).getName() << '\t' << dist << endl;
435 if (output == "lt") { outFile << endl; }
438 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
442 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
448 catch(exception& e) {
449 m->errorOut(e, "DistanceCommand", "driver");
454 /**************************************************************************************************/
455 /////// need to fix to work with calcs and sequencedb
456 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
459 int startTime = time(NULL);
461 string outputString = "";
463 for(int i=startLine;i<endLine;i++){
465 for(int j=0;j<i;j++){
467 if (m->control_pressed) { return 0; }
469 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
470 double dist = distCalculator->getDist();
473 if (output == "column") { outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n'); }
476 if (output == "square") { //make a square column you can convert to square phylip
477 outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
478 outputString += (alignDB.get(j).getName() + ' ' + alignDB.get(i).getName() + ' ' + toString(dist) + '\n');
483 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
484 cout << i << '\t' << (time(NULL) - startTime) << endl;
488 //send results to parent
489 int length = outputString.length();
491 char* buf = new char[length];
\r
492 memcpy(buf, outputString.c_str(), length);
494 MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
500 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
501 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
504 catch(exception& e) {
505 m->errorOut(e, "DistanceCommand", "driverMPI");
509 /**************************************************************************************************/
510 /////// need to fix to work with calcs and sequencedb
511 int DistanceCommand::driverMPI(int startLine, int endLine, string file, long& size){
516 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
518 char* filename = new char[file.length()];
\r
519 memcpy(filename, file.c_str(), file.length());
521 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
524 int startTime = time(NULL);
526 string outputString = "";
529 if((output == "lt") && startLine == 0){ outputString += toString(alignDB.getNumSeqs()) + "\n"; }
531 for(int i=startLine;i<endLine;i++){
533 string name = alignDB.get(i).getName();
534 if (name.length() < 10) { //pad with spaces to make compatible
535 while (name.length() < 10) { name += " "; }
537 outputString += name + "\t";
539 for(int j=0;j<i;j++){
541 if (m->control_pressed) { return 0; }
543 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
544 double dist = distCalculator->getDist();
546 if (output == "lt") { outputString += toString(dist) + "\t"; }
549 if (output == "lt") { outputString += "\n"; }
553 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
554 cout << i << '\t' << (time(NULL) - startTime) << endl;
558 //send results to parent
559 int length = outputString.length();
560 char* buf = new char[length];
\r
561 memcpy(buf, outputString.c_str(), length);
563 MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
564 size += outputString.length();
569 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
570 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
571 MPI_File_close(&outMPI);
575 catch(exception& e) {
576 m->errorOut(e, "DistanceCommand", "driverMPI");
581 /**************************************************************************************************/
582 int DistanceCommand::convertMatrix(string outputFile) {
585 //sort file by first column so the distances for each row are together
586 string outfile = getRootName(outputFile) + "sorted.dist.temp";
589 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
590 string command = "sort -n " + outputFile + " -o " + outfile;
591 system(command.c_str());
592 #else //sort using windows sort
593 string command = "sort " + outputFile + " /O " + outfile;
594 system(command.c_str());
598 //output to new file distance for each row and save positions in file where new row begins
600 openInputFile(outfile, in);
603 openOutputFile(outputFile, out);
605 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
607 out << alignDB.getNumSeqs() << endl;
609 //get first currentRow
610 string first, currentRow, second;
612 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
613 map<string, float>::iterator it;
618 rowDists[first] = 0.00; //distance to yourself is 0.0
621 //openInputFile(outfile, in);
624 if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
626 in >> first >> second >> dist; gobble(in);
628 if (first != currentRow) {
630 out << currentRow << '\t'; //print name
633 for (it = rowDists.begin(); it != rowDists.end(); it++) {
634 out << it->second << '\t';
641 rowDists[first] = 0.00;
642 rowDists[second] = dist;
644 rowDists[second] = dist;
648 out << currentRow << '\t'; //print name
651 for (it = rowDists.begin(); it != rowDists.end(); it++) {
652 out << it->second << '\t';
659 remove(outfile.c_str());
664 catch(exception& e) {
665 m->errorOut(e, "DistanceCommand", "convertMatrix");
669 /**************************************************************************************************/
670 int DistanceCommand::convertToLowerTriangle(string outputFile) {
673 //sort file by first column so the distances for each row are together
674 string outfile = getRootName(outputFile) + "sorted.dist.temp";
677 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
678 string command = "sort -n " + outputFile + " -o " + outfile;
679 system(command.c_str());
680 #else //sort using windows sort
681 string command = "sort " + outputFile + " /O " + outfile;
682 system(command.c_str());
686 //output to new file distance for each row and save positions in file where new row begins
688 openInputFile(outfile, in);
691 openOutputFile(outputFile, out);
693 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
695 out << alignDB.getNumSeqs() << endl;
697 //get first currentRow
698 string first, currentRow, second;
702 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
703 map<string, float>::iterator it;
708 rowDists[first] = 0.00; //distance to yourself is 0.0
711 //openInputFile(outfile, in);
714 if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
716 in >> first >> second >> dist; gobble(in);
718 if (first != currentRow) {
720 out << currentRow << '\t'; //print name
723 for (it = rowDists.begin(); it != rowDists.end(); it++) {
724 if (j >= i) { break; }
725 out << it->second << '\t';
733 rowDists[first] = 0.00;
734 rowDists[second] = dist;
738 rowDists[second] = dist;
742 out << currentRow << '\t'; //print name
745 for (it = rowDists.begin(); it != rowDists.end(); it++) {
746 out << it->second << '\t';
753 remove(outfile.c_str());
758 catch(exception& e) {
759 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
763 /**************************************************************************************************
764 void DistanceCommand::appendFiles(string temp, string filename) {
769 //open output file in append mode
770 openOutputFileAppend(filename, output);
771 openInputFile(temp, input);
773 while(char c = input.get()){
774 if(input.eof()) { break; }
775 else { output << c; }
781 catch(exception& e) {
782 m->errorOut(e, "DistanceCommand", "appendFiles");
786 /**************************************************************************************************/