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()];
\r
260 memcpy(outFilename, filteredFasta.c_str(), filteredFasta.length());
262 char* inFileName = new char[fastafileNames[s].length()];
\r
263 memcpy(inFileName, fastafileNames[s].c_str(), fastafileNames[s].length());
265 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
266 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
271 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
273 if (pid == 0) { //you are the root process
275 MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
278 //send file positions to all processes
279 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
280 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
282 //figure out how many sequences you have to do
283 numSeqsPerProcessor = num / processors;
284 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
285 int startIndex = pid * numSeqsPerProcessor;
288 driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
290 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
293 for(int i = 1; i < processors; i++) {
295 MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
298 }else { //you are a child process
299 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
301 MPIPos.resize(num+1);
302 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
304 //figure out how many sequences you have to align
305 numSeqsPerProcessor = num / processors;
306 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
307 int startIndex = pid * numSeqsPerProcessor;
310 driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
312 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
317 //tell parent you are done.
318 MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
321 MPI_File_close(&outMPI);
322 MPI_File_close(&inMPI);
325 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
328 openInputFile(fastafileNames[s], inFASTA);
329 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
332 lines.push_back(new linePair(0, numFastaSeqs));
334 driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
336 setLines(fastafileNames[s]);
337 createProcessesRunFilter(filter, fastafileNames[s]);
339 rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
342 for(int i=1;i<processors;i++){
343 appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
344 remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
348 if (m->control_pressed) { return 1; }
351 openInputFile(fastafileNames[s], inFASTA);
352 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
355 lines.push_back(new linePair(0, numFastaSeqs));
357 driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
359 if (m->control_pressed) { return 1; }
362 outputNames.push_back(filteredFasta);
367 catch(exception& e) {
368 m->errorOut(e, "FilterSeqsCommand", "filterSequences");
373 /**************************************************************************************/
374 int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<long>& MPIPos) {
376 string outputString = "";
380 for(int i=0;i<num;i++){
382 if (m->control_pressed) { return 0; }
385 int length = MPIPos[start+i+1] - MPIPos[start+i];
386 char* buf4 = new char[length];
387 MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
389 string tempBuf = buf4;
390 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
391 istringstream iss (tempBuf,istringstream::in);
394 Sequence seq(iss); gobble(iss);
396 if (seq.getName() != "") {
397 string align = seq.getAligned();
398 string filterSeq = "";
400 for(int j=0;j<alignmentLength;j++){
401 if(filter[j] == '1'){
402 filterSeq += align[j];
407 outputString += ">" + seq.getName() + "\n" + filterSeq + "\n";
409 if(count % 10 == 0){ //output to file
410 //send results to parent
411 int length = outputString.length();
412 char* buf = new char[length];
\r
413 memcpy(buf, outputString.c_str(), length);
415 MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
422 if((i+1) % 100 == 0){ cout << (i+1) << endl; m->mothurOutJustToLog(toString(i+1) + "\n"); }
425 if(outputString != ""){ //output to file
426 //send results to parent
427 int length = outputString.length();
428 char* buf = new char[length];
\r
429 memcpy(buf, outputString.c_str(), length);
431 MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
436 if((num) % 100 != 0){ cout << (num) << endl; m->mothurOutJustToLog(toString(num) + "\n"); }
440 catch(exception& e) {
441 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
446 /**************************************************************************************/
447 int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* line) {
450 openOutputFile(outputFilename, out);
453 openInputFile(inputFilename, in);
455 in.seekg(line->start);
457 for(int i=0;i<line->num;i++){
459 if (m->control_pressed) { in.close(); out.close(); return 0; }
462 if (seq.getName() != "") {
463 string align = seq.getAligned();
464 string filterSeq = "";
466 for(int j=0;j<alignmentLength;j++){
467 if(filter[j] == '1'){
468 filterSeq += align[j];
472 out << '>' << seq.getName() << endl << filterSeq << endl;
481 catch(exception& e) {
482 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
486 /**************************************************************************************************/
488 int FilterSeqsCommand::createProcessesRunFilter(string F, string filename) {
490 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
495 //loop through and create all the processes you want
496 while (process != processors) {
500 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
503 string filteredFasta = filename + toString(getpid()) + ".temp";
504 driverRunFilter(F, filteredFasta, filename, lines[process]);
506 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
509 //force parent to wait until all the processes are done
510 for (int i=0;i<processors;i++) {
511 int temp = processIDS[i];
518 catch(exception& e) {
519 m->errorOut(e, "FilterSeqsCommand", "createProcessesRunFilter");
523 /**************************************************************************************/
524 string FilterSeqsCommand::createFilter() {
526 string filterString = "";
529 if (soft != 0) { F.setSoft(soft); }
530 if (trump != '*') { F.setTrump(trump); }
532 F.setLength(alignmentLength);
534 if(soft != 0 || isTrue(vertical)){
538 if(hard.compare("") != 0) { F.doHard(hard); }
539 else { F.setFilter(string(alignmentLength, '1')); }
542 if(trump != '*' || isTrue(vertical) || soft != 0){
543 for (int s = 0; s < fastafileNames.size(); s++) {
545 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
548 int pid, numSeqsPerProcessor, num;
554 MPI_Comm_size(MPI_COMM_WORLD, &processors);
555 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
557 char* tempFileName = new char(fastafileNames[s].length());
558 tempFileName = &(fastafileNames[s][0]);
560 MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
562 if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
564 if (pid == 0) { //you are the root process
565 MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
568 //send file positions to all processes
569 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
570 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
572 //figure out how many sequences you have to do
573 numSeqsPerProcessor = num / processors;
574 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
575 int startIndex = pid * numSeqsPerProcessor;
578 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
580 if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
582 }else { //i am the child process
584 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
585 MPIPos.resize(num+1);
587 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
589 //figure out how many sequences you have to align
590 numSeqsPerProcessor = num / processors;
591 if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
592 int startIndex = pid * numSeqsPerProcessor;
595 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
597 if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
600 MPI_File_close(&inMPI);
603 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
606 openInputFile(fastafileNames[s], inFASTA);
607 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
610 numSeqs += numFastaSeqs;
612 lines.push_back(new linePair(0, numFastaSeqs));
614 driverCreateFilter(F, fastafileNames[s], lines[0]);
616 setLines(fastafileNames[s]);
617 createProcessesCreateFilter(F, fastafileNames[s]);
620 if (m->control_pressed) { return filterString; }
623 openInputFile(fastafileNames[s], inFASTA);
624 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
627 numSeqs += numFastaSeqs;
629 lines.push_back(new linePair(0, numFastaSeqs));
631 driverCreateFilter(F, fastafileNames[s], lines[0]);
632 if (m->control_pressed) { return filterString; }
642 int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
645 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
647 if(trump != '*' || isTrue(vertical) || soft != 0){
649 if (pid == 0) { //only one process should output the filter
651 vector<int> temp; temp.resize(alignmentLength+1);
653 //get the frequencies from the child processes
654 for(int i = 0; i < ((processors-1)*5); i++) {
655 MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, MPI_ANY_SOURCE, 2001, MPI_COMM_WORLD, &status);
656 int receiveTag = temp[temp.size()-1]; //child process added a int to the end to indicate what letter count this is for
658 if (receiveTag == Atag) { //you are recieveing the A frequencies
659 for (int k = 0; k < alignmentLength; k++) { F.a[k] += temp[k]; }
660 }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
661 for (int k = 0; k < alignmentLength; k++) { F.t[k] += temp[k]; }
662 }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
663 for (int k = 0; k < alignmentLength; k++) { F.c[k] += temp[k]; }
664 }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
665 for (int k = 0; k < alignmentLength; k++) { F.g[k] += temp[k]; }
666 }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
667 for (int k = 0; k < alignmentLength; k++) { F.gap[k] += temp[k]; }
672 //send my fequency counts
674 int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
676 ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
678 ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
680 ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
681 F.gap.push_back(Gaptag);
682 ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
687 if (pid == 0) { //only one process should output the filter
689 F.setNumSeqs(numSeqs);
691 if(isTrue(vertical) == 1) { F.doVertical(); }
692 if(soft != 0) { F.doSoft(); }
694 filterString = F.getFilter();
697 //send filter string to kids
698 MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
700 //recieve filterString
701 char* tempBuf = new char[alignmentLength];
702 MPI_Bcast(tempBuf, alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
704 filterString = tempBuf;
705 if (filterString.length() > alignmentLength) { filterString = filterString.substr(0, alignmentLength); }
709 MPI_Barrier(MPI_COMM_WORLD);
715 catch(exception& e) {
716 m->errorOut(e, "FilterSeqsCommand", "createFilter");
720 /**************************************************************************************/
721 int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* line) {
725 openInputFile(filename, in);
727 in.seekg(line->start);
729 for(int i=0;i<line->num;i++){
731 if (m->control_pressed) { in.close(); return 1; }
734 if (seq.getName() != "") {
735 if (seq.getAligned().length() != alignmentLength) { m->mothurOut("Sequences are not all the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
737 if(trump != '*'){ F.doTrump(seq); }
738 if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
743 if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); }
747 if((line->num) % 100 != 0){ m->mothurOut(toString(line->num)); m->mothurOutEndLine(); }
753 catch(exception& e) {
754 m->errorOut(e, "FilterSeqsCommand", "driverCreateFilter");
759 /**************************************************************************************/
760 int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<long>& MPIPos) {
765 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
767 for(int i=0;i<num;i++){
769 if (m->control_pressed) { return 0; }
772 int length = MPIPos[start+i+1] - MPIPos[start+i];
774 char* buf4 = new char[length];
775 MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
777 string tempBuf = buf4;
778 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
779 istringstream iss (tempBuf,istringstream::in);
784 if (seq.getAligned().length() != alignmentLength) { cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1); }
786 if(trump != '*'){ F.doTrump(seq); }
787 if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
791 if((i+1) % 100 == 0){ cout << (i+1) << endl; m->mothurOutJustToLog(toString(i+1) + "\n"); }
795 if((num) % 100 != 0){ cout << num << endl; m->mothurOutJustToLog(toString(num) + "\n"); }
799 catch(exception& e) {
800 m->errorOut(e, "FilterSeqsCommand", "MPICreateFilter");
805 /**************************************************************************************************/
807 int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
809 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
814 //loop through and create all the processes you want
815 while (process != processors) {
819 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
822 driverCreateFilter(F, filename, lines[process]);
824 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
827 //force parent to wait until all the processes are done
828 for (int i=0;i<processors;i++) {
829 int temp = processIDS[i];
836 catch(exception& e) {
837 m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
841 /**************************************************************************************************/
843 int FilterSeqsCommand::setLines(string filename) {
846 vector<long int> positions;
850 openInputFile(filename, inFASTA);
853 while(!inFASTA.eof()){
854 input = getline(inFASTA);
856 if (input.length() != 0) {
857 if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
862 int numFastaSeqs = positions.size();
867 //get num bytes in file
868 pFile = fopen (filename.c_str(),"rb");
869 if (pFile==NULL) perror ("Error opening file");
871 fseek (pFile, 0, SEEK_END);
876 numSeqs += numFastaSeqs;
878 int numSeqsPerProcessor = numFastaSeqs / processors;
880 for (int i = 0; i < processors; i++) {
882 long int startPos = positions[ i * numSeqsPerProcessor ];
883 if(i == processors - 1){
884 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
885 bufferSizes.push_back(size - startPos);
887 long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
888 bufferSizes.push_back(myEnd-startPos);
890 lines.push_back(new linePair(startPos, numSeqsPerProcessor));
895 catch(exception& e) {
896 m->errorOut(e, "FilterSeqsCommand", "setLines");
900 /**************************************************************************************/