*/
#include "distancecommand.h"
-#include "ignoregaps.h"
-#include "eachgapdist.h"
-#include "eachgapignore.h"
-#include "onegapdist.h"
-#include "onegapignore.h"
//**********************************************************************************************************************
vector<string> DistanceCommand::setParameters(){
m->openInputFile(fastafile, inFASTA);
alignDB = SequenceDB(inFASTA);
inFASTA.close();
+ m->setFastaFile(fastafile);
}
oldfastafile = validParameter.validFile(parameters, "oldfasta", true);
column = validParameter.validFile(parameters, "column", true);
if (column == "not found") { column = ""; }
else if (column == "not open") { abort = true; }
+ else { m->setColumnFile(column); }
//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"){
if ((column != "") && (oldfastafile != "") && (output != "column")) { m->mothurOut("You have provided column and oldfasta, indicating you want to append distances to your column file. Your output must be in column format to do so."); m->mothurOutEndLine(); abort=true; }
if ((output != "column") && (output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are column, lt and square. I will use column."); m->mothurOutEndLine(); output = "column"; }
-
- ValidCalculators validCalculator;
-
- 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(); }
- else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
- else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
- }
- }
- }else {
- for (int i=0; i<Estimators.size(); i++) {
- if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
- if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
- else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
- else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
- }
- }
- }
}
int numSeqs = alignDB.getNumSeqs();
cutoff += 0.005;
+ if (!alignDB.sameLength()) { m->mothurOut("[ERROR]: your sequences are not the same length, aborting."); m->mothurOutEndLine(); return 0; }
+
string outputFile;
if (output == "lt") { //does the user want lower triangle phylip formatted file
outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
- remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
+ m->mothurRemove(outputFile); outputTypes["phylip"].push_back(outputFile);
//output numSeqs to phylip formatted dist file
}else if (output == "column") { //user wants column format
rename(column.c_str(), tempcolumn.c_str());
}
- remove(outputFile.c_str());
+ m->mothurRemove(outputFile);
}else { //assume square
outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
- remove(outputFile.c_str());
+ m->mothurRemove(outputFile);
outputTypes["phylip"].push_back(outputFile);
}
//do your part
string outputMyPart;
- unsigned long int mySize;
+ unsigned long long mySize;
if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
else { driverMPI(start, end, outputFile, mySize, output); }
//wait on chidren
for(int b = 1; b < processors; b++) {
- unsigned long int fileSize;
+ unsigned long long fileSize;
if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator; return 0; }
MPI_File_close(&outMPI);
}else { //you are a child process
//do your part
- unsigned long int size;
+ unsigned long long size;
if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//if you don't need to fork anything
if(processors == 1){
if (output != "square") { driver(0, numSeqs, outputFile, cutoff); }
else { driver(0, numSeqs, outputFile, "square"); }
}else{ //you have multiple processors
- unsigned long int numDists = 0;
+ unsigned long long numDists = 0;
if (output == "square") {
numDists = numSeqs * numSeqs;
createProcesses(outputFile);
}
- #else
+ //#else
//ifstream inFASTA;
- if (output != "square") { driver(0, numSeqs, outputFile, cutoff); }
- else { driver(0, numSeqs, outputFile, "square"); }
- #endif
+ //if (output != "square") { driver(0, numSeqs, outputFile, cutoff); }
+ //else { driver(0, numSeqs, outputFile, "square"); }
+ //#endif
#endif
- if (m->control_pressed) { outputTypes.clear(); delete distCalculator; remove(outputFile.c_str()); return 0; }
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFile); return 0; }
#ifdef USE_MPI
MPI_Comm_rank(MPI_COMM_WORLD, &pid);
if (outputFile == column) {
string tempcolumn = column + ".old";
m->appendFiles(tempcolumn, outputFile);
- remove(tempcolumn.c_str());
+ m->mothurRemove(tempcolumn);
}else{
m->appendFiles(outputFile, column);
- remove(outputFile.c_str());
+ m->mothurRemove(outputFile);
outputFile = column;
}
if (outputDir != "") {
string newOutputName = outputDir + m->getSimpleName(outputFile);
rename(outputFile.c_str(), newOutputName.c_str());
- remove(outputFile.c_str());
+ m->mothurRemove(outputFile);
outputFile = newOutputName;
}
}
}
#endif
- if (m->control_pressed) { outputTypes.clear(); delete distCalculator; remove(outputFile.c_str()); return 0; }
-
- delete distCalculator;
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFile); return 0; }
//set phylip file as new current phylipfile
string current = "";
int temp = processIDS[i];
wait(&temp);
}
+#else
+ //////////////////////////////////////////////////////////////////////////////////////////////////////
+ //Windows version shared memory, so be careful when passing variables through the distanceData struct.
+ //Above fork() will clone, so memory is separate, but that's not the case with windows,
+ //that's why the distance calculator was moved inside of the driver to make separate copies.
+ //////////////////////////////////////////////////////////////////////////////////////////////////////
+
+ vector<distanceData*> pDataArray; //[processors-1];
+ DWORD dwThreadIdArray[processors-1];
+ HANDLE hThreadArray[processors-1];
+
+ //Create processor-1 worker threads.
+ for( int i=0; i<processors-1; i++ ){
+
+ // Allocate memory for thread data.
+ distanceData* tempDist = new distanceData(lines[i+1].start, lines[i+1].end, (filename + toString(i) + ".temp"), cutoff, alignDB, Estimators, m, output, numNewFasta, countends);
+ pDataArray.push_back(tempDist);
+ processIDS.push_back(i);
+
+ //MyDistThreadFunction is in header. It must be global or static to work with the threads.
+ //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
+ hThreadArray[i] = CreateThread(NULL, 0, MyDistThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
+ }
+
+ //do your part
+ if (output != "square") { driver(lines[0].start, lines[0].end, filename, cutoff); }
+ else { driver(lines[0].start, lines[0].end, filename, "square"); }
+
+ //Wait until all threads have terminated.
+ WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
+
+ //Close all thread handles and free memory allocations.
+ for(int i=0; i < pDataArray.size(); i++){
+ CloseHandle(hThreadArray[i]);
+ delete pDataArray[i];
+ }
+#endif
//append and remove temp files
for (int i=0;i<processIDS.size();i++) {
m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
- remove((filename + toString(processIDS[i]) + ".temp").c_str());
+ m->mothurRemove((filename + toString(processIDS[i]) + ".temp"));
}
-#endif
+
}
catch(exception& e) {
m->errorOut(e, "DistanceCommand", "createProcesses");
exit(1);
}
}
-
/**************************************************************************************************/
/////// need to fix to work with calcs and sequencedb
int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
try {
-
+ ValidCalculators validCalculator;
+ Dist* distCalculator;
+ 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(); }
+ else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
+ }
+ }
+ }else {
+ for (int i=0; i<Estimators.size(); i++) {
+ if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
+ if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
+ else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
+ }
+ }
+ }
+
int startTime = time(NULL);
//column file
}
for(int j=0;j<i;j++){
- if (m->control_pressed) { outFile.close(); return 0; }
+ if (m->control_pressed) { delete distCalculator; outFile.close(); return 0; }
//if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
//the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
outFile.close();
+ delete distCalculator;
return 1;
}
/////// need to fix to work with calcs and sequencedb
int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
try {
-
+ ValidCalculators validCalculator;
+ Dist* distCalculator;
+ 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(); }
+ else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
+ }
+ }
+ }else {
+ for (int i=0; i<Estimators.size(); i++) {
+ if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
+ if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
+ else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
+ }
+ }
+ }
+
int startTime = time(NULL);
//column file
for(int j=0;j<alignDB.getNumSeqs();j++){
- if (m->control_pressed) { outFile.close(); return 0; }
+ if (m->control_pressed) { delete distCalculator; outFile.close(); return 0; }
distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
double dist = distCalculator->getDist();
m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
outFile.close();
+ delete distCalculator;
return 1;
}
/////// need to fix to work with calcs and sequencedb
int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
try {
+
+ ValidCalculators validCalculator;
+ Dist* distCalculator;
+ 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(); }
+ else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
+ }
+ }
+ }else {
+ for (int i=0; i<Estimators.size(); i++) {
+ if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
+ if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
+ else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
+ }
+ }
+ }
+
+
MPI_Status status;
int startTime = time(NULL);
for(int j=0;j<i;j++){
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { delete distCalculator; return 0; }
//if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
//the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
}
//m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ delete distCalculator;
return 1;
}
catch(exception& e) {
}
/**************************************************************************************************/
/////// need to fix to work with calcs and sequencedb
-int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long long& size){
try {
+ ValidCalculators validCalculator;
+ Dist* distCalculator;
+ 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(); }
+ else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
+ }
+ }
+ }else {
+ for (int i=0; i<Estimators.size(); i++) {
+ if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
+ if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
+ else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
+ }
+ }
+ }
+
+
MPI_Status status;
MPI_File outMPI;
for(int j=0;j<i;j++){
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { delete distCalculator; return 0; }
distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
double dist = distCalculator->getDist();
//m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
MPI_File_close(&outMPI);
+ delete distCalculator;
return 1;
}
}
/**************************************************************************************************/
/////// need to fix to work with calcs and sequencedb
-int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long long& size, string square){
try {
+ ValidCalculators validCalculator;
+ Dist* distCalculator;
+ 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(); }
+ else if (Estimators[i] == "eachgap") { distCalculator = new eachGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapDist(); }
+ }
+ }
+ }else {
+ for (int i=0; i<Estimators.size(); i++) {
+ if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
+ if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
+ else if (Estimators[i] == "eachgap"){ distCalculator = new eachGapIgnoreTermGapDist(); }
+ else if (Estimators[i] == "onegap") { distCalculator = new oneGapIgnoreTermGapDist(); }
+ }
+ }
+ }
+
MPI_Status status;
MPI_File outMPI;
for(int j=0;j<alignDB.getNumSeqs();j++){
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { delete distCalculator; return 0; }
distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
double dist = distCalculator->getDist();
//m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
MPI_File_close(&outMPI);
-
+ delete distCalculator;
return 1;
}
catch(exception& e) {
//m->openInputFile(outfile, in);
while(!in.eof()) {
- if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
+ if (m->control_pressed) { in.close(); m->mothurRemove(outfile); out.close(); return 0; }
in >> first >> second >> dist; m->gobble(in);
in.close();
out.close();
- remove(outfile.c_str());
+ m->mothurRemove(outfile);
return 1;
//m->openInputFile(outfile, in);
while(!in.eof()) {
- if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
+ if (m->control_pressed) { in.close(); m->mothurRemove(outfile); out.close(); return 0; }
in >> first >> second >> dist; m->gobble(in);
in.close();
out.close();
- remove(outfile.c_str());
+ m->mothurRemove(outfile);
return 1;
string name1, name2;
float dist;
while (!inDist.eof()) {
- if (m->control_pressed) { inDist.close(); outDist.close(); remove(outputFile.c_str()); return good; }
+ if (m->control_pressed) { inDist.close(); outDist.close(); m->mothurRemove(outputFile); return good; }
inDist >> name1 >> name2 >> dist; m->gobble(inDist);
outDist.close();
if (good) {
- remove(column.c_str());
+ m->mothurRemove(column);
rename(outputFile.c_str(), column.c_str());
}else{
- remove(outputFile.c_str()); //temp file is bad because file mismatch above
+ m->mothurRemove(outputFile); //temp file is bad because file mismatch above
}
return good;
projects / mothur.git / blobdiff