2 * filterseqscommand.cpp
5 * Created by Thomas Ryabin on 5/4/09.
6 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
10 #include "filterseqscommand.h"
11 #include "sequence.hpp"
13 /**************************************************************************************/
15 FilterSeqsCommand::FilterSeqsCommand(string option) {
20 //allow user to run help
21 if(option == "help") { help(); abort = true; }
24 //valid paramters for this command
25 string Array[] = {"fasta", "trump", "soft", "hard", "vertical", "outputdir","inputdir", "processors"};
26 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
28 OptionParser parser(option);
29 map<string,string> parameters = parser.getParameters();
31 ValidParameters validParameter;
32 map<string,string>::iterator it;
34 //check to make sure all parameters are valid for command
35 for (it = parameters.begin(); it != parameters.end(); it++) {
36 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
39 //if the user changes the input directory command factory will send this info to us in the output parameter
40 string inputDir = validParameter.validFile(parameters, "inputdir", false);
41 if (inputDir == "not found"){ inputDir = ""; }
44 it = parameters.find("fasta");
45 //user has given a template file
46 if(it != parameters.end()){
47 path = hasPath(it->second);
48 //if the user has not given a path then, add inputdir. else leave path alone.
49 if (path == "") { parameters["fasta"] = inputDir + it->second; }
52 it = parameters.find("hard");
53 //user has given a template file
54 if(it != parameters.end()){
55 path = hasPath(it->second);
56 //if the user has not given a path then, add inputdir. else leave path alone.
57 if (path == "") { parameters["hard"] = inputDir + it->second; }
61 //check for required parameters
62 fasta = validParameter.validFile(parameters, "fasta", false);
63 if (fasta == "not found") { m->mothurOut("fasta is a required parameter for the filter.seqs command."); m->mothurOutEndLine(); abort = true; }
65 splitAtDash(fasta, fastafileNames);
67 //go through files and make sure they are good, if not, then disregard them
68 for (int i = 0; i < fastafileNames.size(); i++) {
70 string path = hasPath(fastafileNames[i]);
71 //if the user has not given a path then, add inputdir. else leave path alone.
72 if (path == "") { fastafileNames[i] = inputDir + fastafileNames[i]; }
77 ableToOpen = openInputFile(fastafileNames[i], in);
78 if (ableToOpen == 1) {
79 m->mothurOut(fastafileNames[i] + " will be disregarded."); m->mothurOutEndLine();
80 //erase from file list
81 fastafileNames.erase(fastafileNames.begin()+i);
84 string simpleName = getSimpleName(fastafileNames[i]);
85 filterFileName += simpleName.substr(0, simpleName.find_first_of('.'));
90 //make sure there is at least one valid file left
91 if (fastafileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
95 //if the user changes the output directory command factory will send this info to us in the output parameter
96 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
98 outputDir += hasPath(fastafileNames[0]); //if user entered a file with a path then preserve it
101 //check for optional parameter and set defaults
102 // ...at some point should added some additional type checking...
105 temp = validParameter.validFile(parameters, "trump", false); if (temp == "not found") { temp = "*"; }
108 temp = validParameter.validFile(parameters, "soft", false); if (temp == "not found") { soft = 0; }
109 else { soft = (float)atoi(temp.c_str()) / 100.0; }
111 temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
112 convert(temp, processors);
114 hard = validParameter.validFile(parameters, "hard", true); if (hard == "not found") { hard = ""; }
115 else if (hard == "not open") { abort = true; }
117 vertical = validParameter.validFile(parameters, "vertical", false); if (vertical == "not found") { vertical = "T"; }
124 catch(exception& e) {
125 m->errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
130 //**********************************************************************************************************************
132 void FilterSeqsCommand::help(){
135 m->mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
136 m->mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
137 m->mothurOut("The fasta parameter is required. You may enter several fasta files to build the filter from and filter, by separating their names with -'s.\n");
138 m->mothurOut("For example: fasta=abrecovery.fasta-amazon.fasta \n");
139 m->mothurOut("The trump parameter .... The default is ...\n");
140 m->mothurOut("The soft parameter .... The default is ....\n");
141 m->mothurOut("The hard parameter .... The default is ....\n");
142 m->mothurOut("The vertical parameter .... The default is T.\n");
143 m->mothurOut("The filter.seqs command should be in the following format: \n");
144 m->mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
145 m->mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
146 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
149 catch(exception& e) {
150 m->errorOut(e, "FilterSeqsCommand", "help");
155 /**************************************************************************************/
157 int FilterSeqsCommand::execute() {
160 if (abort == true) { return 0; }
163 openInputFile(fastafileNames[0], inFASTA);
165 Sequence testSeq(inFASTA);
166 alignmentLength = testSeq.getAlignLength();
169 ////////////create filter/////////////////
170 m->mothurOut("Creating Filter... "); m->mothurOutEndLine();
172 filter = createFilter();
174 m->mothurOutEndLine(); m->mothurOutEndLine();
176 if (m->control_pressed) { return 0; }
180 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
182 if (pid == 0) { //only one process should output the filter
187 string filterFile = outputDir + filterFileName + ".filter";
188 openOutputFile(filterFile, outFilter);
189 outFilter << filter << endl;
191 outputNames.push_back(filterFile);
197 ////////////run filter/////////////////
199 m->mothurOut("Running Filter... "); m->mothurOutEndLine();
203 m->mothurOutEndLine(); m->mothurOutEndLine();
205 int filteredLength = 0;
206 for(int i=0;i<alignmentLength;i++){
207 if(filter[i] == '1'){ filteredLength++; }
210 if (m->control_pressed) { for(int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
213 m->mothurOutEndLine();
214 m->mothurOut("Length of filtered alignment: " + toString(filteredLength)); m->mothurOutEndLine();
215 m->mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); m->mothurOutEndLine();
216 m->mothurOut("Length of the original alignment: " + toString(alignmentLength)); m->mothurOutEndLine();
217 m->mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); m->mothurOutEndLine();
220 m->mothurOutEndLine();
221 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
222 for(int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
223 m->mothurOutEndLine();
228 catch(exception& e) {
229 m->errorOut(e, "FilterSeqsCommand", "execute");
233 /**************************************************************************************/
234 int FilterSeqsCommand::filterSequences() {
239 for (int s = 0; s < fastafileNames.size(); s++) {
241 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
243 string filteredFasta = outputDir + getRootName(getSimpleName(fastafileNames[s])) + "filter.fasta";
245 int pid, start, end, numSeqsPerProcessor, num;
250 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
251 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
256 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
257 int inMode=MPI_MODE_RDONLY;
259 //char* outFilename = new char[filteredFasta.length()];
260 //memcpy(outFilename, filteredFasta.c_str(), filteredFasta.length());
262 char outFilename[1024];
263 strcpy(outFilename, filteredFasta.c_str());
265 //char* inFileName = new char[fastafileNames[s].length()];
266 //memcpy(inFileName, fastafileNames[s].c_str(), fastafileNames[s].length());
268 char inFileName[1024];
269 strcpy(inFileName, fastafileNames[s].c_str());
271 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
272 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
275 //delete outFilename;
277 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
279 if (pid == 0) { //you are the root process
281 MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
284 //send file positions to all processes
285 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
286 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
288 //figure out how many sequences you have to do
289 numSeqsPerProcessor = num / processors;
290 int startIndex = pid * numSeqsPerProcessor;
291 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
295 driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
297 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
300 for(int i = 1; i < processors; i++) {
302 MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
305 }else { //you are a child process
306 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
308 MPIPos.resize(num+1);
309 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
311 //figure out how many sequences you have to align
312 numSeqsPerProcessor = num / processors;
313 int startIndex = pid * numSeqsPerProcessor;
314 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
318 driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
320 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
325 //tell parent you are done.
326 MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
329 MPI_File_close(&outMPI);
330 MPI_File_close(&inMPI);
333 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
336 openInputFile(fastafileNames[s], inFASTA);
337 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
340 lines.push_back(new linePair(0, numFastaSeqs));
342 driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
344 setLines(fastafileNames[s]);
345 createProcessesRunFilter(filter, fastafileNames[s]);
347 rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
350 for(int i=1;i<processors;i++){
351 appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
352 remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
356 if (m->control_pressed) { return 1; }
359 openInputFile(fastafileNames[s], inFASTA);
360 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
363 lines.push_back(new linePair(0, numFastaSeqs));
365 driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
367 if (m->control_pressed) { return 1; }
370 outputNames.push_back(filteredFasta);
375 catch(exception& e) {
376 m->errorOut(e, "FilterSeqsCommand", "filterSequences");
381 /**************************************************************************************/
382 int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<long>& MPIPos) {
384 string outputString = "";
388 for(int i=0;i<num;i++){
390 if (m->control_pressed) { return 0; }
393 int length = MPIPos[start+i+1] - MPIPos[start+i];
394 char* buf4 = new char[length];
395 MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
397 string tempBuf = buf4;
398 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
399 istringstream iss (tempBuf,istringstream::in);
402 Sequence seq(iss); gobble(iss);
404 if (seq.getName() != "") {
405 string align = seq.getAligned();
406 string filterSeq = "";
408 for(int j=0;j<alignmentLength;j++){
409 if(filter[j] == '1'){
410 filterSeq += align[j];
415 outputString += ">" + seq.getName() + "\n" + filterSeq + "\n";
417 if(count % 10 == 0){ //output to file
418 //send results to parent
419 int length = outputString.length();
420 char* buf = new char[length];
421 memcpy(buf, outputString.c_str(), length);
423 MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
430 if((i+1) % 100 == 0){ cout << (i+1) << endl; m->mothurOutJustToLog(toString(i+1) + "\n"); }
433 if(outputString != ""){ //output to file
434 //send results to parent
435 int length = outputString.length();
436 char* buf = new char[length];
437 memcpy(buf, outputString.c_str(), length);
439 MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
444 if((num) % 100 != 0){ cout << (num) << endl; m->mothurOutJustToLog(toString(num) + "\n"); }
448 catch(exception& e) {
449 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
454 /**************************************************************************************/
455 int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* line) {
458 openOutputFile(outputFilename, out);
461 openInputFile(inputFilename, in);
463 in.seekg(line->start);
465 for(int i=0;i<line->num;i++){
467 if (m->control_pressed) { in.close(); out.close(); return 0; }
470 if (seq.getName() != "") {
471 string align = seq.getAligned();
472 string filterSeq = "";
474 for(int j=0;j<alignmentLength;j++){
475 if(filter[j] == '1'){
476 filterSeq += align[j];
480 out << '>' << seq.getName() << endl << filterSeq << endl;
489 catch(exception& e) {
490 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
494 /**************************************************************************************************/
496 int FilterSeqsCommand::createProcessesRunFilter(string F, string filename) {
498 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
503 //loop through and create all the processes you want
504 while (process != processors) {
508 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
511 string filteredFasta = filename + toString(getpid()) + ".temp";
512 driverRunFilter(F, filteredFasta, filename, lines[process]);
514 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
517 //force parent to wait until all the processes are done
518 for (int i=0;i<processors;i++) {
519 int temp = processIDS[i];
526 catch(exception& e) {
527 m->errorOut(e, "FilterSeqsCommand", "createProcessesRunFilter");
531 /**************************************************************************************/
532 string FilterSeqsCommand::createFilter() {
534 string filterString = "";
537 if (soft != 0) { F.setSoft(soft); }
538 if (trump != '*') { F.setTrump(trump); }
540 F.setLength(alignmentLength);
542 if(soft != 0 || isTrue(vertical)){
546 if(hard.compare("") != 0) { F.doHard(hard); }
547 else { F.setFilter(string(alignmentLength, '1')); }
550 if(trump != '*' || isTrue(vertical) || soft != 0){
551 for (int s = 0; s < fastafileNames.size(); s++) {
553 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
556 int pid, numSeqsPerProcessor, num;
562 MPI_Comm_size(MPI_COMM_WORLD, &processors);
563 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
565 //char* tempFileName = new char(fastafileNames[s].length());
566 //tempFileName = &(fastafileNames[s][0]);
568 char tempFileName[1024];
569 strcpy(tempFileName, fastafileNames[s].c_str());
571 MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
573 if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
575 if (pid == 0) { //you are the root process
576 MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
579 //send file positions to all processes
580 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
581 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
583 //figure out how many sequences you have to do
584 numSeqsPerProcessor = num / processors;
585 int startIndex = pid * numSeqsPerProcessor;
586 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
590 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
592 if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
594 }else { //i am the child process
596 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
597 MPIPos.resize(num+1);
599 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
601 //figure out how many sequences you have to align
602 numSeqsPerProcessor = num / processors;
603 int startIndex = pid * numSeqsPerProcessor;
604 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
608 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
610 if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
613 MPI_File_close(&inMPI);
616 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
619 openInputFile(fastafileNames[s], inFASTA);
620 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
623 numSeqs += numFastaSeqs;
625 lines.push_back(new linePair(0, numFastaSeqs));
627 driverCreateFilter(F, fastafileNames[s], lines[0]);
629 setLines(fastafileNames[s]);
630 createProcessesCreateFilter(F, fastafileNames[s]);
633 if (m->control_pressed) { return filterString; }
636 openInputFile(fastafileNames[s], inFASTA);
637 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
640 numSeqs += numFastaSeqs;
642 lines.push_back(new linePair(0, numFastaSeqs));
644 driverCreateFilter(F, fastafileNames[s], lines[0]);
645 if (m->control_pressed) { return filterString; }
655 int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
658 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
660 if(trump != '*' || isTrue(vertical) || soft != 0){
662 if (pid == 0) { //only one process should output the filter
664 vector<int> temp; temp.resize(alignmentLength+1);
666 //get the frequencies from the child processes
667 for(int i = 0; i < ((processors-1)*5); i++) {
668 MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, MPI_ANY_SOURCE, 2001, MPI_COMM_WORLD, &status);
669 int receiveTag = temp[temp.size()-1]; //child process added a int to the end to indicate what letter count this is for
671 if (receiveTag == Atag) { //you are recieveing the A frequencies
672 for (int k = 0; k < alignmentLength; k++) { F.a[k] += temp[k]; }
673 }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
674 for (int k = 0; k < alignmentLength; k++) { F.t[k] += temp[k]; }
675 }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
676 for (int k = 0; k < alignmentLength; k++) { F.c[k] += temp[k]; }
677 }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
678 for (int k = 0; k < alignmentLength; k++) { F.g[k] += temp[k]; }
679 }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
680 for (int k = 0; k < alignmentLength; k++) { F.gap[k] += temp[k]; }
685 //send my fequency counts
687 int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
689 ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
691 ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
693 ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
694 F.gap.push_back(Gaptag);
695 ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
700 if (pid == 0) { //only one process should output the filter
702 F.setNumSeqs(numSeqs);
704 if(isTrue(vertical) == 1) { F.doVertical(); }
705 if(soft != 0) { F.doSoft(); }
707 filterString = F.getFilter();
710 //send filter string to kids
711 MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
713 //recieve filterString
714 char* tempBuf = new char[alignmentLength];
715 MPI_Bcast(tempBuf, alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
717 filterString = tempBuf;
718 if (filterString.length() > alignmentLength) { filterString = filterString.substr(0, alignmentLength); }
722 MPI_Barrier(MPI_COMM_WORLD);
728 catch(exception& e) {
729 m->errorOut(e, "FilterSeqsCommand", "createFilter");
733 /**************************************************************************************/
734 int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* line) {
738 openInputFile(filename, in);
740 in.seekg(line->start);
742 for(int i=0;i<line->num;i++){
744 if (m->control_pressed) { in.close(); return 1; }
747 if (seq.getName() != "") {
748 if (seq.getAligned().length() != alignmentLength) { m->mothurOut("Sequences are not all the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
750 if(trump != '*'){ F.doTrump(seq); }
751 if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
756 if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); }
760 if((line->num) % 100 != 0){ m->mothurOut(toString(line->num)); m->mothurOutEndLine(); }
766 catch(exception& e) {
767 m->errorOut(e, "FilterSeqsCommand", "driverCreateFilter");
772 /**************************************************************************************/
773 int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<long>& MPIPos) {
778 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
780 for(int i=0;i<num;i++){
782 if (m->control_pressed) { return 0; }
785 int length = MPIPos[start+i+1] - MPIPos[start+i];
787 char* buf4 = new char[length];
788 MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
790 string tempBuf = buf4;
791 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
792 istringstream iss (tempBuf,istringstream::in);
797 if (seq.getAligned().length() != alignmentLength) { cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1); }
799 if(trump != '*'){ F.doTrump(seq); }
800 if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
804 if((i+1) % 100 == 0){ cout << (i+1) << endl; m->mothurOutJustToLog(toString(i+1) + "\n"); }
808 if((num) % 100 != 0){ cout << num << endl; m->mothurOutJustToLog(toString(num) + "\n"); }
812 catch(exception& e) {
813 m->errorOut(e, "FilterSeqsCommand", "MPICreateFilter");
818 /**************************************************************************************************/
820 int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
822 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
827 //loop through and create all the processes you want
828 while (process != processors) {
832 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
835 driverCreateFilter(F, filename, lines[process]);
837 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
840 //force parent to wait until all the processes are done
841 for (int i=0;i<processors;i++) {
842 int temp = processIDS[i];
849 catch(exception& e) {
850 m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
854 /**************************************************************************************************/
856 int FilterSeqsCommand::setLines(string filename) {
859 vector<long int> positions;
863 openInputFile(filename, inFASTA);
866 while(!inFASTA.eof()){
867 input = getline(inFASTA);
869 if (input.length() != 0) {
870 if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
875 int numFastaSeqs = positions.size();
880 //get num bytes in file
881 pFile = fopen (filename.c_str(),"rb");
882 if (pFile==NULL) perror ("Error opening file");
884 fseek (pFile, 0, SEEK_END);
889 numSeqs += numFastaSeqs;
891 int numSeqsPerProcessor = numFastaSeqs / processors;
893 for (int i = 0; i < processors; i++) {
895 long int startPos = positions[ i * numSeqsPerProcessor ];
896 if(i == processors - 1){
897 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
898 bufferSizes.push_back(size - startPos);
900 long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
901 bufferSizes.push_back(myEnd-startPos);
903 lines.push_back(new linePair(startPos, numSeqsPerProcessor));
908 catch(exception& e) {
909 m->errorOut(e, "FilterSeqsCommand", "setLines");
913 /**************************************************************************************/