OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
- ValidParameters validParameter;
+ ValidParameters validParameter("filter.seqs");
map<string,string>::iterator it;
//check to make sure all parameters are valid for command
if (path == "") { fastafileNames[i] = inputDir + fastafileNames[i]; }
}
- int ableToOpen;
ifstream in;
- ableToOpen = openInputFile(fastafileNames[i], in);
+ int ableToOpen = openInputFile(fastafileNames[i], in, "noerror");
+
+ //if you can't open it, try default location
+ if (ableToOpen == 1) {
+ if (m->getDefaultPath() != "") { //default path is set
+ string tryPath = m->getDefaultPath() + getSimpleName(fastafileNames[i]);
+ m->mothurOut("Unable to open " + fastafileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
+ ableToOpen = openInputFile(tryPath, in, "noerror");
+ fastafileNames[i] = tryPath;
+ }
+ }
+ in.close();
+
if (ableToOpen == 1) {
- m->mothurOut(fastafileNames[i] + " will be disregarded."); m->mothurOutEndLine();
+ m->mothurOut("Unable to open " + fastafileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
//erase from file list
fastafileNames.erase(fastafileNames.begin()+i);
i--;
// ...at some point should added some additional type checking...
string temp;
+ hard = validParameter.validFile(parameters, "hard", true); if (hard == "not found") { hard = ""; }
+ else if (hard == "not open") { hard = ""; abort = true; }
+
temp = validParameter.validFile(parameters, "trump", false); if (temp == "not found") { temp = "*"; }
trump = temp[0];
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
convert(temp, processors);
- hard = validParameter.validFile(parameters, "hard", true); if (hard == "not found") { hard = ""; }
- else if (hard == "not open") { abort = true; }
-
- vertical = validParameter.validFile(parameters, "vertical", false); if (vertical == "not found") { vertical = "T"; }
+ vertical = validParameter.validFile(parameters, "vertical", false);
+ if (vertical == "not found") {
+ if ((hard == "") && (trump == '*') && (soft == 0)) { vertical = "T"; } //you have not given a hard file or set the trump char.
+ else { vertical = "F"; }
+ }
numSeqs = 0;
-
}
}
m->mothurOut("For example: fasta=abrecovery.fasta-amazon.fasta \n");
m->mothurOut("The trump parameter .... The default is ...\n");
m->mothurOut("The soft parameter .... The default is ....\n");
- m->mothurOut("The hard parameter .... The default is ....\n");
- m->mothurOut("The vertical parameter .... The default is T.\n");
+ m->mothurOut("The hard parameter allows you to enter a file containing the filter you want to use.\n");
+ m->mothurOut("The vertical parameter removes columns where all sequences contain a gap character. The default is T.\n");
m->mothurOut("The filter.seqs command should be in the following format: \n");
- m->mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
- m->mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
+ m->mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump) \n");
+ m->mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=.).\n");
m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
}
inFASTA.close();
////////////create filter/////////////////
+ m->mothurOut("Creating Filter... "); m->mothurOutEndLine();
filter = createFilter();
+ m->mothurOutEndLine(); m->mothurOutEndLine();
+
if (m->control_pressed) { return 0; }
+ #ifdef USE_MPI
+ int pid;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid);
+
+ if (pid == 0) { //only one process should output the filter
+ #endif
+
ofstream outFilter;
string filterFile = outputDir + filterFileName + ".filter";
outFilter.close();
outputNames.push_back(filterFile);
+ #ifdef USE_MPI
+ }
+ #endif
////////////run filter/////////////////
+ m->mothurOut("Running Filter... "); m->mothurOutEndLine();
+
filterSequences();
-
+
+ m->mothurOutEndLine(); m->mothurOutEndLine();
+
int filteredLength = 0;
for(int i=0;i<alignmentLength;i++){
if(filter[i] == '1'){ filteredLength++; }
/**************************************************************************************/
int FilterSeqsCommand::filterSequences() {
try {
-
+
+ numSeqs = 0;
+
for (int s = 0; s < fastafileNames.size(); s++) {
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
string filteredFasta = outputDir + getRootName(getSimpleName(fastafileNames[s])) + "filter.fasta";
#ifdef USE_MPI
- int pid, start, end;
+ int pid, start, end, numSeqsPerProcessor, num;
int tag = 2001;
+ vector<unsigned long int>MPIPos;
MPI_Status status;
MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
MPI_File outMPI;
+ MPI_File tempMPI;
MPI_File inMPI;
int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
int inMode=MPI_MODE_RDONLY;
- char outFilename[filteredFasta.length()];
+ char outFilename[1024];
strcpy(outFilename, filteredFasta.c_str());
-
- char inFileName[fastafileNames[s].length()];
+
+ char inFileName[1024];
strcpy(inFileName, fastafileNames[s].c_str());
MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
-
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
+
if (pid == 0) { //you are the root process
- setLines(fastafileNames[s]);
-
- for (int j = 0; j < lines.size(); j++) { //each process
- if (j != 0) { //don't send to yourself
- MPI_Send(&lines[j]->start, 1, MPI_INT, j, tag, MPI_COMM_WORLD); //start position in file
- MPI_Send(&bufferSizes[j], 1, MPI_INT, j, tag, MPI_COMM_WORLD); //how bytes for the read
- }
+ MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
+ numSeqs += num;
+
+ //send file positions to all processes
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
}
- //read your peice of file
- char buf[bufferSizes[0]];
- MPI_File_read_at(inMPI, lines[0]->start, buf, bufferSizes[0], MPI_CHAR, &status);
- istringstream iss (buf,istringstream::in);
+ //figure out how many sequences you have to do
+ numSeqsPerProcessor = num / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
+
//do your part
- driverMPIRun(iss, outMPI);
+ driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
//wait on chidren
for(int i = 1; i < processors; i++) {
}
}else { //you are a child process
- //receive your section of file
- int startPos, numLines, bufferSize;
- MPI_Recv(&startPos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
- MPI_Recv(&bufferSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
-
- //read your peice of file
- char buf2[bufferSize];
- MPI_File_read_at(inMPI, startPos, buf2, bufferSize, MPI_CHAR, &status);
- istringstream iss (buf2,istringstream::in);
+ MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ MPIPos.resize(num+1);
+ numSeqs += num;
+ MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = num / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
+
+
+ //align your part
+ driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
- //do your part
- driverMPIRun(iss, outMPI);
-
char buf[4];
strcpy(buf, "done");
MPI_File_close(&outMPI);
MPI_File_close(&inMPI);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
ifstream inFASTA;
+ int numFastaSeqs;
openInputFile(fastafileNames[s], inFASTA);
- int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ getNumSeqs(inFASTA, numFastaSeqs);
inFASTA.close();
lines.push_back(new linePair(0, numFastaSeqs));
+ numSeqs += numFastaSeqs;
+
driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
}else{
- setLines(fastafileNames[s]);
+ setLines(fastafileNames[s]);
createProcessesRunFilter(filter, fastafileNames[s]);
-
+
rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
//append fasta files
for(int i=1;i<processors;i++){
- appendAlignFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
+ appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
}
}
if (m->control_pressed) { return 1; }
#else
ifstream inFASTA;
+ int numFastaSeqs;
openInputFile(fastafileNames[s], inFASTA);
- int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ getNumSeqs(inFASTA, numFastaSeqs);
inFASTA.close();
lines.push_back(new linePair(0, numFastaSeqs));
+ numSeqs += numFastaSeqs;
+
driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
if (m->control_pressed) { return 1; }
exit(1);
}
}
+#ifdef USE_MPI
/**************************************************************************************/
-int FilterSeqsCommand::driverMPIRun(istringstream& in, MPI_File& outMPI) {
+int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
try {
string outputString = "";
int count = 0;
MPI_Status status;
- while (!in.eof()) {
+ for(int i=0;i<num;i++){
+
+ if (m->control_pressed) { return 0; }
+
+ //read next sequence
+ int length = MPIPos[start+i+1] - MPIPos[start+i];
+ char* buf4 = new char[length];
+ MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
- Sequence seq(in); gobble(in);
+ string tempBuf = buf4;
+ if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
+ istringstream iss (tempBuf,istringstream::in);
+ delete buf4;
+
+ Sequence seq(iss); gobble(iss);
if (seq.getName() != "") {
string align = seq.getAligned();
if(count % 10 == 0){ //output to file
//send results to parent
int length = outputString.length();
- char buf[length];
- strcpy(buf, outputString.c_str());
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
outputString = "";
+ delete buf;
}
}
+
+ if((i+1) % 100 == 0){ cout << (i+1) << endl; m->mothurOutJustToLog(toString(i+1) + "\n"); }
}
if(outputString != ""){ //output to file
//send results to parent
int length = outputString.length();
- char buf[length];
- strcpy(buf, outputString.c_str());
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
outputString = "";
+ delete buf;
}
-
+
+ if((num) % 100 != 0){ cout << (num) << endl; m->mothurOutJustToLog(toString(num) + "\n"); }
return 0;
}
exit(1);
}
}
+#endif
/**************************************************************************************/
int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* line) {
try {
in.seekg(line->start);
- for(int i=0;i<line->numSeqs;i++){
+ for(int i=0;i<line->num;i++){
if (m->control_pressed) { in.close(); out.close(); return 0; }
out << '>' << seq.getName() << endl << filterSeq << endl;
}
gobble(in);
+
+ //report progress
+ if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); }
}
+
+ //report progress
+ if((line->num) % 100 != 0){ m->mothurOut(toString(line->num)); m->mothurOutEndLine(); }
+
out.close();
in.close();
/**************************************************************************************/
string FilterSeqsCommand::createFilter() {
try {
- string filterString = "";
-
+ string filterString = "";
Filters F;
if (soft != 0) { F.setSoft(soft); }
F.setLength(alignmentLength);
- if(soft != 0 || isTrue(vertical)){
+ if(trump != '*' || isTrue(vertical) || soft != 0){
F.initialize();
}
else { F.setFilter(string(alignmentLength, '1')); }
numSeqs = 0;
-
if(trump != '*' || isTrue(vertical) || soft != 0){
for (int s = 0; s < fastafileNames.size(); s++) {
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
+
#ifdef USE_MPI
- int pid, rc, ierr;
- int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
+ int pid, numSeqsPerProcessor, num;
int tag = 2001;
+ vector<unsigned long int> MPIPos;
MPI_Status status;
- MPI_File in;
- rc = MPI_Comm_size(MPI_COMM_WORLD, &processors);
- rc = MPI_Comm_rank(MPI_COMM_WORLD, &pid);
+ MPI_File inMPI;
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid);
- char tempFileName[fastafileNames[s].length()];
+ //char* tempFileName = new char(fastafileNames[s].length());
+ //tempFileName = &(fastafileNames[s][0]);
+
+ char tempFileName[1024];
strcpy(tempFileName, fastafileNames[s].c_str());
- cout << pid << " tempFileName " << tempFileName << endl;
- MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &in); //comm, filename, mode, info, filepointer
- cout << pid << " here" << endl;
+
+ MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
+
if (pid == 0) { //you are the root process
- setLines(fastafileNames[s]);
- cout << pid << " after setlines" << endl;
- for (int j = 0; j < lines.size(); j++) { //each process
- if (j != 0) { //don't send to yourself
- MPI_Send(&lines[j]->start, 1, MPI_INT, j, tag, MPI_COMM_WORLD); //start position in file
- MPI_Send(&bufferSizes[j], 1, MPI_INT, j, tag, MPI_COMM_WORLD); //how bytes for the read
- }
+ MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
+ numSeqs += num;
+
+ //send file positions to all processes
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
}
- cout << pid << " done sending" << endl;
- char buf[bufferSizes[0]];
- MPI_File_read_at(in, 0, buf, bufferSizes[0], MPI_CHAR, &status);
- cout << pid << " done reading" << endl;
- string tempBuf = buf;
- if (tempBuf.length() > bufferSizes[0]) { tempBuf = tempBuf.substr(0, bufferSizes[0]); }
-
- MPICreateFilter(F, tempBuf);
- cout << pid << "done with mpi create filter " << endl;
- if (m->control_pressed) { MPI_File_close(&in); return filterString; }
-
- vector<int> temp; temp.resize(alignmentLength+1);
+
+ //figure out how many sequences you have to do
+ numSeqsPerProcessor = num / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
- //get the frequencies from the child processes
- for(int i = 0; i < ((processors-1)*5); i++) {
- MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, MPI_ANY_SOURCE, tag, MPI_COMM_WORLD, &status);
- int receiveTag = temp[temp.size()-1]; //child process added a int to the end to indicate what letter count this is for
-
- if (receiveTag == Atag) { //you are recieveing the A frequencies
- for (int k = 0; k < alignmentLength; k++) { F.a[k] += temp[k]; }
- }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
- for (int k = 0; k < alignmentLength; k++) { F.t[k] += temp[k]; }
- }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
- for (int k = 0; k < alignmentLength; k++) { F.c[k] += temp[k]; }
- }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
- for (int k = 0; k < alignmentLength; k++) { F.g[k] += temp[k]; }
- }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
- for (int k = 0; k < alignmentLength; k++) { F.gap[k] += temp[k]; }
- }
- }
-
+
+ //do your part
+ MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
+ if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
+
}else { //i am the child process
- cout << pid << endl;
- int startPos, bufferSize;
- ierr = MPI_Recv(&startPos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
- ierr = MPI_Recv(&bufferSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
- cout << pid << '\t' << startPos << '\t' << bufferSize << endl;
- //send freqs
- char buf2[bufferSize];
- MPI_File_read_at( in, startPos, buf2, bufferSize, MPI_CHAR, &status);
-
- string tempBuf = buf2;
- if (tempBuf.length() > bufferSize) { tempBuf = tempBuf.substr(0, bufferSize); }
-
- MPICreateFilter(F, tempBuf);
- cout << pid << "done with mpi create filter " << endl;
- if (m->control_pressed) { MPI_File_close(&in); return filterString; }
-
- //send my fequency counts
- F.a.push_back(Atag);
- int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, tag, MPI_COMM_WORLD);
- F.t.push_back(Ttag);
- ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, tag, MPI_COMM_WORLD);
- F.c.push_back(Ctag);
- ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, tag, MPI_COMM_WORLD);
- F.g.push_back(Gtag);
- ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, tag, MPI_COMM_WORLD);
- F.gap.push_back(Gaptag);
- ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, tag, MPI_COMM_WORLD);
+ MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ MPIPos.resize(num+1);
+ numSeqs += num;
+ MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = num / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
+
+
+ //do your part
+ MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
}
- MPI_Barrier(MPI_COMM_WORLD);
- MPI_File_close(&in);
+ MPI_File_close(&inMPI);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
ifstream inFASTA;
+ int numFastaSeqs;
openInputFile(fastafileNames[s], inFASTA);
- int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ getNumSeqs(inFASTA, numFastaSeqs);
inFASTA.close();
numSeqs += numFastaSeqs;
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
lines.push_back(new linePair(0, numFastaSeqs));
driverCreateFilter(F, fastafileNames[s], lines[0]);
if (m->control_pressed) { return filterString; }
#else
ifstream inFASTA;
+ int numFastaSeqs;
openInputFile(fastafileNames[s], inFASTA);
- int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ getNumSeqs(inFASTA, numFastaSeqs);
inFASTA.close();
numSeqs += numFastaSeqs;
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
lines.push_back(new linePair(0, numFastaSeqs));
driverCreateFilter(F, fastafileNames[s], lines[0]);
}
}
+
+#ifdef USE_MPI
+ int pid;
+ int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
+ MPI_Status status;
+
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid);
+
+ if(trump != '*' || isTrue(vertical) || soft != 0){
+
+ if (pid == 0) { //only one process should output the filter
+
+ vector<int> temp; temp.resize(alignmentLength+1);
+
+ //get the frequencies from the child processes
+ for(int i = 1; i < processors; i++) {
+
+ for (int j = 0; j < 5; j++) {
+
+ MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, i, 2001, MPI_COMM_WORLD, &status);
+ int receiveTag = temp[temp.size()-1]; //child process added a int to the end to indicate what letter count this is for
+
+ if (receiveTag == Atag) { //you are recieveing the A frequencies
+ for (int k = 0; k < alignmentLength; k++) { F.a[k] += temp[k]; }
+ }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
+ for (int k = 0; k < alignmentLength; k++) { F.t[k] += temp[k]; }
+ }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
+ for (int k = 0; k < alignmentLength; k++) { F.c[k] += temp[k]; }
+ }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
+ for (int k = 0; k < alignmentLength; k++) { F.g[k] += temp[k]; }
+ }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
+ for (int k = 0; k < alignmentLength; k++) { F.gap[k] += temp[k]; }
+ }
+ }
+ }
+ }else{
+
+ //send my fequency counts
+ F.a.push_back(Atag);
+ int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
+ F.t.push_back(Ttag);
+ ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
+ F.c.push_back(Ctag);
+ ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
+ F.g.push_back(Gtag);
+ ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
+ F.gap.push_back(Gaptag);
+ ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
+ }
+
+ }
+
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
+
+ if (pid == 0) { //only one process should output the filter
+#endif
F.setNumSeqs(numSeqs);
if(isTrue(vertical) == 1) { F.doVertical(); }
if(soft != 0) { F.doSoft(); }
filterString = F.getFilter();
-
+
+#ifdef USE_MPI
+ //send filter string to kids
+ //for(int i = 1; i < processors; i++) {
+ // MPI_Send(&filterString[0], alignmentLength, MPI_CHAR, i, 2001, MPI_COMM_WORLD);
+ //}
+ MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
+ }else{
+ //recieve filterString
+ char* tempBuf = new char[alignmentLength];
+ //MPI_Recv(&tempBuf[0], alignmentLength, MPI_CHAR, 0, 2001, MPI_COMM_WORLD, &status);
+ MPI_Bcast(tempBuf, alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
+
+ filterString = tempBuf;
+ if (filterString.length() > alignmentLength) { filterString = filterString.substr(0, alignmentLength); }
+ delete tempBuf;
+ }
+
+ MPI_Barrier(MPI_COMM_WORLD);
+#endif
+
return filterString;
}
catch(exception& e) {
in.seekg(line->start);
- for(int i=0;i<line->numSeqs;i++){
+ for(int i=0;i<line->num;i++){
if (m->control_pressed) { in.close(); return 1; }
}
//report progress
- if((line->numSeqs) % 100 != 0){ m->mothurOut(toString(line->numSeqs)); m->mothurOutEndLine(); }
+ if((line->num) % 100 != 0){ m->mothurOut(toString(line->num)); m->mothurOutEndLine(); }
in.close();
exit(1);
}
}
+#ifdef USE_MPI
/**************************************************************************************/
-int FilterSeqsCommand::MPICreateFilter(Filters& F, string input) {
+int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<unsigned long int>& MPIPos) {
try {
- vector<string> seqStrings;
- parseBuffer(input, seqStrings);
+ MPI_Status status;
+ int pid;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
- for(int i=0;i<seqStrings.size();i++){
+ for(int i=0;i<num;i++){
- if (seqStrings[i].length() != alignmentLength) { cout << i << '\t' << seqStrings[i].length() << "Sequences are not all the same length, please correct." << endl; m->control_pressed = true; }
-
- if (m->control_pressed) { return 1; }
+ if (m->control_pressed) { return 0; }
- Sequence seq("", seqStrings[i]);
+ //read next sequence
+ int length = MPIPos[start+i+1] - MPIPos[start+i];
+
+ char* buf4 = new char[length];
+ MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
+
+ string tempBuf = buf4;
+ if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
+ istringstream iss (tempBuf,istringstream::in);
+ delete buf4;
+
+ Sequence seq(iss);
+
+ if (seq.getAligned().length() != alignmentLength) { cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1); }
if(trump != '*'){ F.doTrump(seq); }
if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
cout.flush();
//report progress
- if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); }
+ if((i+1) % 100 == 0){ cout << (i+1) << endl; m->mothurOutJustToLog(toString(i+1) + "\n"); }
}
//report progress
- if((seqStrings.size()) % 100 != 0){ m->mothurOut(toString(seqStrings.size())); m->mothurOutEndLine(); }
+ if((num) % 100 != 0){ cout << num << endl; m->mothurOutJustToLog(toString(num) + "\n"); }
return 0;
}
exit(1);
}
}
-
+#endif
/**************************************************************************************************/
int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
//loop through and create all the processes you want
while (process != processors) {
- int pid = vfork();
+ int pid = fork();
if (pid > 0) {
processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
process++;
}else if (pid == 0){
driverCreateFilter(F, filename, lines[process]);
+
+ //write out filter counts to file
+ filename += toString(getpid()) + "filterValues.temp";
+ ofstream out;
+ openOutputFile(filename, out);
+
+ for (int k = 0; k < alignmentLength; k++) { out << F.a[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.t[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.g[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.c[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.gap[k] << '\t'; } out << endl;
+
+ out.close();
+
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
wait(&temp);
}
+ //parent reads in and combine Filter info
+ for (int i = 0; i < processIDS.size(); i++) {
+ string tempFilename = filename + toString(processIDS[i]) + "filterValues.temp";
+ ifstream in;
+ openInputFile(tempFilename, in);
+
+ int temp;
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.a[k] += temp; } gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.t[k] += temp; } gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.g[k] += temp; } gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.c[k] += temp; } gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.gap[k] += temp; } gobble(in);
+
+ in.close();
+ remove(tempFilename.c_str());
+ }
+
return exitCommand;
#endif
}
int FilterSeqsCommand::setLines(string filename) {
try {
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- vector<long int> positions;
+
+ vector<unsigned long int> positions;
bufferSizes.clear();
ifstream inFASTA;
input = getline(inFASTA);
if (input.length() != 0) {
- if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
+ if(input[0] == '>'){ unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
}
}
inFASTA.close();
int numFastaSeqs = positions.size();
FILE * pFile;
- long size;
+ unsigned long int size;
//get num bytes in file
pFile = fopen (filename.c_str(),"rb");
for (int i = 0; i < processors; i++) {
- long int startPos = positions[ i * numSeqsPerProcessor ];
+ unsigned long int startPos = positions[ i * numSeqsPerProcessor ];
if(i == processors - 1){
numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
bufferSizes.push_back(size - startPos);
}else{
- long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
+ unsigned long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
bufferSizes.push_back(myEnd-startPos);
}
lines.push_back(new linePair(startPos, numSeqsPerProcessor));
exit(1);
}
}
-/**************************************************************************************************/
-int FilterSeqsCommand::parseBuffer(string file, vector<string>& seqs) {
- try {
- istringstream iss (file); //,istringstream::in
- string name, seqstring;
-
- while (!iss.eof()) {
-
- if (m->control_pressed) { return 0; }
-
- Sequence seq(iss); gobble(iss);
-
- if (seq.getName() != "") {
- seqs.push_back(seq.getAligned());
- }
- }
-
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "FilterSeqsCommand", "parseBuffer");
- exit(1);
- }
-}
/**************************************************************************************/
projects / mothur.git / blobdiff