it2 = parameters.find("fasta");
//user has given a template file
if(it2 != parameters.end()){
- path = hasPath(it2->second);
+ path = m->hasPath(it2->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["fasta"] = inputDir + it2->second; }
}
it2 = parameters.find("oldfasta");
//user has given a template file
if(it2 != parameters.end()){
- path = hasPath(it2->second);
+ path = m->hasPath(it2->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["oldfasta"] = inputDir + it2->second; }
}
it2 = parameters.find("column");
//user has given a template file
if(it2 != parameters.end()){
- path = hasPath(it2->second);
+ path = m->hasPath(it2->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["column"] = inputDir + it2->second; }
}
else if (fastafile == "not open") { abort = true; }
else{
ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
+ m->openInputFile(fastafile, inFASTA);
alignDB = SequenceDB(inFASTA);
inFASTA.close();
}
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
outputDir = "";
- outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
+ outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
}
//check for optional parameter and set defaults
else {
if (calc == "default") { calc = "onegap"; }
}
- splitAtDash(calc, Estimators);
+ m->splitAtDash(calc, Estimators);
string temp;
temp = validParameter.validFile(parameters, "countends", false); if(temp == "not found"){ temp = "T"; }
ValidCalculators validCalculator;
- if (isTrue(countends) == true) {
+ if (m->isTrue(countends) == true) {
for (int i=0; i<Estimators.size(); i++) {
if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
string outputFile;
if (output == "lt") { //does the user want lower triangle phylip formatted file
- outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "phylip.dist";
+ outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
remove(outputFile.c_str());
//output numSeqs to phylip formatted dist file
}else if (output == "column") { //user wants column format
- outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "dist";
+ outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
//so we don't accidentally overwrite
if (outputFile == column) {
remove(outputFile.c_str());
}else { //assume square
- outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "square.dist";
+ outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
remove(outputFile.c_str());
}
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
//each process gets where it should start and stop in the file
- start = int (sqrt(float(pid)/float(processors)) * numSeqs);
- end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
+ if (output != "square") {
+ start = int (sqrt(float(pid)/float(processors)) * numSeqs);
+ end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
+ }else{
+ start = int ((float(pid)/float(processors)) * numSeqs);
+ end = int ((float(pid+1)/float(processors)) * numSeqs);
+ }
- if (output != "lt") {
+ if (output == "column") {
MPI_File outMPI;
int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
//do your part
string outputMyPart;
- driverMPI(start, end, outMPI, cutoff);
+
+ driverMPI(start, end, outMPI, cutoff);
if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
}
}else { //you are a child process
//do your part
- driverMPI(start, end, outMPI, cutoff);
+ driverMPI(start, end, outMPI, cutoff);
if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
//do your part
string outputMyPart;
- long mySize;
- driverMPI(start, end, outputFile, mySize);
+ unsigned long int mySize;
+
+ if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
+ else { driverMPI(start, end, outputFile, mySize, output); }
if (m->control_pressed) { delete distCalculator; return 0; }
//wait on chidren
for(int b = 1; b < processors; b++) {
- long fileSize;
+ unsigned long int fileSize;
if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
MPI_File_close(&outMPI);
}else { //you are a child process
//do your part
- long size;
- driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size);
+ unsigned long int size;
+ if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
+ else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
if (m->control_pressed) { delete distCalculator; return 0; }
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//if you don't need to fork anything
if(processors == 1){
- driver(0, numSeqs, outputFile, cutoff);
+ if (output != "square") { driver(0, numSeqs, outputFile, cutoff); }
+ else { driver(0, numSeqs, outputFile, "square"); }
}else{ //you have multiple processors
for (int i = 0; i < processors; i++) {
lines.push_back(new linePair());
- lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
- lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
+ if (output != "square") {
+ lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
+ lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
+ }else{
+ lines[i]->start = int ((float(i)/float(processors)) * numSeqs);
+ lines[i]->end = int ((float(i+1)/float(processors)) * numSeqs);
+ }
}
createProcesses(outputFile);
//append and remove temp files
for (; it != processIDS.end(); it++) {
- appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
+ m->appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
remove((outputFile + toString(it->second) + ".temp").c_str());
}
}
#else
- ifstream inFASTA;
- driver(0, numSeqs, outputFile, cutoff);
+ //ifstream inFASTA;
+ if (output != "square") { driver(0, numSeqs, outputFile, cutoff); }
+ else { driver(0, numSeqs, outputFile, "square"); }
#endif
#endif
if (pid == 0) { //only one process should output to screen
#endif
- if (output == "square") { convertMatrix(outputFile); }
+ //if (output == "square") { convertMatrix(outputFile); }
ifstream fileHandle;
fileHandle.open(outputFile.c_str());
if(fileHandle) {
- gobble(fileHandle);
+ m->gobble(fileHandle);
if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff."); m->mothurOutEndLine(); }
}
//we had to rename the column file so we didnt overwrite above, but we want to keep old name
if (outputFile == column) {
string tempcolumn = column + ".old";
- appendFiles(tempcolumn, outputFile);
+ m->appendFiles(tempcolumn, outputFile);
remove(tempcolumn.c_str());
}else{
- appendFiles(outputFile, column);
+ m->appendFiles(outputFile, column);
remove(outputFile.c_str());
outputFile = column;
}
processIDS[lines[process]->end] = pid; //create map from line number to pid so you can append files in correct order later
process++;
}else if (pid == 0){
- driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
+ if (output != "square") { driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff); }
+ else { driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", "square"); }
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
}
if (output == "lt") { outFile << dist << '\t'; }
-
- if (output == "square") { //make a square column you can convert to square phylip
- outFile << alignDB.get(i).getName() << '\t' << alignDB.get(j).getName() << '\t' << dist << endl;
- outFile << alignDB.get(j).getName() << '\t' << alignDB.get(i).getName() << '\t' << dist << endl;
- }
-
}
if (output == "lt") { outFile << endl; }
exit(1);
}
}
+/**************************************************************************************************/
+/////// need to fix to work with calcs and sequencedb
+int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
+ try {
+
+ int startTime = time(NULL);
+
+ //column file
+ ofstream outFile(dFileName.c_str(), ios::trunc);
+ outFile.setf(ios::fixed, ios::showpoint);
+ outFile << setprecision(4);
+
+ if(startLine == 0){ outFile << alignDB.getNumSeqs() << endl; }
+
+ for(int i=startLine;i<endLine;i++){
+
+ string name = alignDB.get(i).getName();
+ //pad with spaces to make compatible
+ if (name.length() < 10) { while (name.length() < 10) { name += " "; } }
+
+ outFile << name << '\t';
+
+ for(int j=0;j<alignDB.getNumSeqs();j++){
+
+ if (m->control_pressed) { outFile.close(); return 0; }
+
+ distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
+ double dist = distCalculator->getDist();
+
+ outFile << dist << '\t';
+ }
+
+ outFile << endl;
+
+ if(i % 100 == 0){
+ m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ }
+
+ }
+ m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+
+ outFile.close();
+
+ return 1;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "DistanceCommand", "driver");
+ exit(1);
+ }
+}
#ifdef USE_MPI
/**************************************************************************************************/
/////// need to fix to work with calcs and sequencedb
double dist = distCalculator->getDist();
if(dist <= cutoff){
- if (output == "column") { outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n'); }
- }
-
- if (output == "square") { //make a square column you can convert to square phylip
- outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
- outputString += (alignDB.get(j).getName() + ' ' + alignDB.get(i).getName() + ' ' + toString(dist) + '\n');
+ outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
}
}
}
/**************************************************************************************************/
/////// need to fix to work with calcs and sequencedb
-int DistanceCommand::driverMPI(int startLine, int endLine, string file, long& size){
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
try {
MPI_Status status;
string outputString = "";
size = 0;
- if((output == "lt") && startLine == 0){ outputString += toString(alignDB.getNumSeqs()) + "\n"; }
+ if(startLine == 0){ outputString += toString(alignDB.getNumSeqs()) + "\n"; }
for(int i=startLine;i<endLine;i++){
- if(output == "lt") {
- string name = alignDB.get(i).getName();
- if (name.length() < 10) { //pad with spaces to make compatible
- while (name.length() < 10) { name += " "; }
- }
- outputString += name + "\t";
+
+ string name = alignDB.get(i).getName();
+ if (name.length() < 10) { //pad with spaces to make compatible
+ while (name.length() < 10) { name += " "; }
}
+ outputString += name + "\t";
+
for(int j=0;j<i;j++){
if (m->control_pressed) { return 0; }
distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
double dist = distCalculator->getDist();
- if (output == "lt") { outputString += toString(dist) + "\t"; }
+ outputString += toString(dist) + "\t";
}
- if (output == "lt") { outputString += "\n"; }
+ outputString += "\n";
if(i % 100 == 0){
exit(1);
}
}
-#endif
/**************************************************************************************************/
+/////// need to fix to work with calcs and sequencedb
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
+ try {
+ MPI_Status status;
+
+ MPI_File outMPI;
+ int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+
+ //char* filename = new char[file.length()];
+ //memcpy(filename, file.c_str(), file.length());
+
+ char filename[1024];
+ strcpy(filename, file.c_str());
+
+ MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
+ //delete filename;
+
+ int startTime = time(NULL);
+
+ string outputString = "";
+ size = 0;
+
+ if(startLine == 0){ outputString += toString(alignDB.getNumSeqs()) + "\n"; }
+
+ for(int i=startLine;i<endLine;i++){
+
+ string name = alignDB.get(i).getName();
+ if (name.length() < 10) { //pad with spaces to make compatible
+ while (name.length() < 10) { name += " "; }
+ }
+ outputString += name + "\t";
+
+ for(int j=0;j<alignDB.getNumSeqs();j++){
+
+ if (m->control_pressed) { return 0; }
+
+ distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
+ double dist = distCalculator->getDist();
+
+ outputString += toString(dist) + "\t";
+ }
+
+ outputString += "\n";
+
+
+ if(i % 100 == 0){
+ //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ cout << i << '\t' << (time(NULL) - startTime) << endl;
+ }
+
+
+ //send results to parent
+ int length = outputString.length();
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
+
+ MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
+ size += outputString.length();
+ outputString = "";
+ delete buf;
+ }
+
+ //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ MPI_File_close(&outMPI);
+
+ return 1;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "DistanceCommand", "driverMPI");
+ exit(1);
+ }
+}
+#endif
+/**************************************************************************************************
int DistanceCommand::convertMatrix(string outputFile) {
try{
//sort file by first column so the distances for each row are together
- string outfile = getRootName(outputFile) + "sorted.dist.temp";
+ string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
//use the unix sort
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//output to new file distance for each row and save positions in file where new row begins
ifstream in;
- openInputFile(outfile, in);
+ m->openInputFile(outfile, in);
ofstream out;
- openOutputFile(outputFile, out);
+ m->openOutputFile(outputFile, out);
out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
rowDists[first] = 0.00; //distance to yourself is 0.0
in.seekg(0);
- //openInputFile(outfile, in);
+ //m->openInputFile(outfile, in);
while(!in.eof()) {
if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
- in >> first >> second >> dist; gobble(in);
+ in >> first >> second >> dist; m->gobble(in);
if (first != currentRow) {
//print out last row
exit(1);
}
}
-/**************************************************************************************************/
+/**************************************************************************************************
int DistanceCommand::convertToLowerTriangle(string outputFile) {
try{
//sort file by first column so the distances for each row are together
- string outfile = getRootName(outputFile) + "sorted.dist.temp";
+ string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
//use the unix sort
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//output to new file distance for each row and save positions in file where new row begins
ifstream in;
- openInputFile(outfile, in);
+ m->openInputFile(outfile, in);
ofstream out;
- openOutputFile(outputFile, out);
+ m->openOutputFile(outputFile, out);
out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
rowDists[first] = 0.00; //distance to yourself is 0.0
in.seekg(0);
- //openInputFile(outfile, in);
+ //m->openInputFile(outfile, in);
while(!in.eof()) {
if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
- in >> first >> second >> dist; gobble(in);
+ in >> first >> second >> dist; m->gobble(in);
if (first != currentRow) {
//print out last row
//make sure the 2 fasta files have the same alignment length
ifstream in;
- openInputFile(fastafile, in);
+ m->openInputFile(fastafile, in);
int fastaAlignLength = 0;
if (in) {
Sequence tempIn(in);
in.close();
ifstream in2;
- openInputFile(oldfastafile, in2);
+ m->openInputFile(oldfastafile, in2);
int oldfastaAlignLength = 0;
if (in2) {
Sequence tempIn2(in2);
set<string> namesOldFasta;
ifstream inFasta;
- openInputFile(oldfastafile, inFasta);
+ m->openInputFile(oldfastafile, inFasta);
while (!inFasta.eof()) {
if (m->control_pressed) { inFasta.close(); return good; }
alignDB.push_back(temp); //add to DB
}
- gobble(inFasta);
+ m->gobble(inFasta);
}
inFasta.close();
//read through the column file checking names and removing distances above the cutoff
ifstream inDist;
- openInputFile(column, inDist);
+ m->openInputFile(column, inDist);
ofstream outDist;
string outputFile = column + ".temp";
- openOutputFile(outputFile, outDist);
+ m->openOutputFile(outputFile, outDist);
string name1, name2;
float dist;
while (!inDist.eof()) {
if (m->control_pressed) { inDist.close(); outDist.close(); remove(outputFile.c_str()); return good; }
- inDist >> name1 >> name2 >> dist; gobble(inDist);
+ inDist >> name1 >> name2 >> dist; m->gobble(inDist);
//both names are in fasta file and distance is below cutoff
if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) { good = false; break; }
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "appendFiles");
- exit(1);
- }
-}
-
-/**************************************************************************************************
-void DistanceCommand::appendFiles(string temp, string filename) {
- try{
- ofstream output;
- ifstream input;
-
- //open output file in append mode
- openOutputFileAppend(filename, output);
- openInputFile(temp, input);
-
- while(char c = input.get()){
- if(input.eof()) { break; }
- else { output << c; }
- }
-
- input.close();
- output.close();
- }
- catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "appendFiles");
+ m->errorOut(e, "DistanceCommand", "m->appendFiles");
exit(1);
}
}
/**************************************************************************************************/
+
+
projects / mothur.git / blobdiff