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");
}
#ifdef USE_MPI
int pid, start, end, numSeqsPerProcessor, num;
int tag = 2001;
- vector<long>MPIPos;
+ 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
-
+ cout << pid << "is in create filter " << endl;
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
numSeqs += num;
//send file positions to all processes
- MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
- MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
+ 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);
+ }
//figure out how many sequences you have to do
numSeqsPerProcessor = num / processors;
- if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
+
//do your part
driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
}
}else { //you are a child process
- MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
- numSeqs += num;
+ MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
MPIPos.resize(num+1);
- MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
+ 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;
- if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
+
//align your part
driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
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]);
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; }
}
#ifdef USE_MPI
/**************************************************************************************/
-int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<long>& MPIPos) {
+int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
try {
string outputString = "";
int count = 0;
//read next sequence
int length = MPIPos[start+i+1] - MPIPos[start+i];
- char buf4[length];
+ 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); gobble(iss);
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(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"); }
F.setLength(alignmentLength);
- if(soft != 0 || isTrue(vertical)){
+ if(trump != '*' || isTrue(vertical) || soft != 0){
F.initialize();
}
#ifdef USE_MPI
int pid, numSeqsPerProcessor, num;
int tag = 2001;
- vector<long> MPIPos;
+ vector<unsigned long int> MPIPos;
MPI_Status status;
MPI_File inMPI;
MPI_Comm_size(MPI_COMM_WORLD, &processors);
MPI_Comm_rank(MPI_COMM_WORLD, &pid);
- char* tempFileName = new char(fastafileNames[s].length());
- tempFileName = &(fastafileNames[s][0]);
+ //char* tempFileName = new char(fastafileNames[s].length());
+ //tempFileName = &(fastafileNames[s][0]);
+
+ char tempFileName[1024];
+ strcpy(tempFileName, fastafileNames[s].c_str());
MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
numSeqs += num;
//send file positions to all processes
- MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
- MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
+ 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);
+ }
//figure out how many sequences you have to do
numSeqsPerProcessor = num / processors;
- if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
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; }
}else { //i am the child process
-
- MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
+ MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
MPIPos.resize(num+1);
numSeqs += num;
- MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
+ 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;
- if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
+
//do your part
MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
}
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;
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;
MPI_Status status;
MPI_Comm_rank(MPI_COMM_WORLD, &pid);
- 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 = 0; i < ((processors-1)*5); i++) {
- MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, MPI_ANY_SOURCE, 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);
+ 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);
#ifdef USE_MPI
//send filter string to kids
- MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
+ //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[alignmentLength];
+ 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) {
}
#ifdef USE_MPI
/**************************************************************************************/
-int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<long>& MPIPos) {
+int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<unsigned long int>& MPIPos) {
try {
MPI_Status status;
//read next sequence
int length = MPIPos[start+i+1] - MPIPos[start+i];
- char buf4[length];
+ 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); }
int FilterSeqsCommand::setLines(string filename) {
try {
- 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));