*/
#include "distancecommand.h"
-#include "ignoregaps.h"
-#include "eachgapdist.h"
-#include "eachgapignore.h"
-#include "onegapdist.h"
-#include "onegapignore.h"
//**********************************************************************************************************************
-vector<string> DistanceCommand::getValidParameters(){
+vector<string> DistanceCommand::setParameters(){
try {
- string Array[] = {"fasta","oldfasta","column", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir","compress"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ CommandParameter pcolumn("column", "InputTypes", "", "", "none", "none", "OldFastaColumn","column",false,false); parameters.push_back(pcolumn);
+ CommandParameter poldfasta("oldfasta", "InputTypes", "", "", "none", "none", "OldFastaColumn","",false,false); parameters.push_back(poldfasta);
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","phylip-column",false,true, true); parameters.push_back(pfasta);
+ CommandParameter poutput("output", "Multiple", "column-lt-square-phylip", "column", "", "", "","phylip-column",false,false, true); parameters.push_back(poutput);
+ CommandParameter pcalc("calc", "Multiple", "nogaps-eachgap-onegap", "onegap", "", "", "","",false,false); parameters.push_back(pcalc);
+ CommandParameter pcountends("countends", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pcountends);
+ CommandParameter pcompress("compress", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pcompress);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false, true); parameters.push_back(pprocessors);
+ CommandParameter pcutoff("cutoff", "Number", "", "1.0", "", "", "","",false,false, true); parameters.push_back(pcutoff);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
return myArray;
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "getValidParameters");
+ m->errorOut(e, "DistanceCommand", "setParameters");
exit(1);
}
}
//**********************************************************************************************************************
-DistanceCommand::DistanceCommand(){
+string DistanceCommand::getHelpString(){
try {
- //initialize outputTypes
- vector<string> tempOutNames;
- outputTypes["phylip"] = tempOutNames;
- outputTypes["column"] = tempOutNames;
+ string helpString = "";
+ helpString += "The dist.seqs command reads a file containing sequences and creates a distance file.\n";
+ helpString += "The dist.seqs command parameters are fasta, oldfasta, column, calc, countends, output, compress, cutoff and processors. \n";
+ helpString += "The fasta parameter is required, unless you have a valid current fasta file.\n";
+ helpString += "The oldfasta and column parameters allow you to append the distances calculated to the column file.\n";
+ helpString += "The calc parameter allows you to specify the method of calculating the distances. Your options are: nogaps, onegap or eachgap. The default is onegap.\n";
+ helpString += "The countends parameter allows you to specify whether to include terminal gaps in distance. Your options are: T or F. The default is T.\n";
+ helpString += "The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n";
+ helpString += "The output parameter allows you to specify format of your distance matrix. Options are column, lt, and square. The default is column.\n";
+ helpString += "The processors parameter allows you to specify number of processors to use. The default is 1.\n";
+ helpString += "The compress parameter allows you to indicate that you want the resulting distance file compressed. The default is false.\n";
+ helpString += "The dist.seqs command should be in the following format: \n";
+ helpString += "dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n";
+ helpString += "Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n";
+ return helpString;
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "DistanceCommand");
+ m->errorOut(e, "DistanceCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
-vector<string> DistanceCommand::getRequiredParameters(){
- try {
- string Array[] = {"fasta"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "ChopSeqsCommand", "getRequiredParameters");
- exit(1);
- }
+string DistanceCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "phylip") { pattern = "[filename],[outputtag],dist"; }
+ else if (type == "column") { pattern = "[filename],dist"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "DistanceCommand", "getOutputPattern");
+ exit(1);
+ }
}
//**********************************************************************************************************************
-vector<string> DistanceCommand::getRequiredFiles(){
+DistanceCommand::DistanceCommand(){
try {
- vector<string> myArray;
- return myArray;
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["phylip"] = tempOutNames;
+ outputTypes["column"] = tempOutNames;
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "getRequiredFiles");
+ m->errorOut(e, "DistanceCommand", "DistanceCommand");
exit(1);
}
}
//**********************************************************************************************************************
DistanceCommand::DistanceCommand(string option) {
try {
- abort = false;
+ abort = false; calledHelp = false;
Estimators.clear();
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string Array[] = {"fasta","oldfasta","column", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir","compress"};
-
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string, string> parameters = parser.getParameters();
//check for required parameters
fastafile = validParameter.validFile(parameters, "fasta", true);
- if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the dist.seqs command."); m->mothurOutEndLine(); abort = true; }
- else if (fastafile == "not open") { abort = true; }
+ if (fastafile == "not found") {
+ fastafile = m->getFastaFile();
+ if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine();
+ ifstream inFASTA;
+ m->openInputFile(fastafile, inFASTA);
+ alignDB = SequenceDB(inFASTA);
+ inFASTA.close();
+ }else { m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }else if (fastafile == "not open") { abort = true; }
else{
ifstream inFASTA;
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"){
convert(temp, countends);
temp = validParameter.validFile(parameters, "cutoff", false); if(temp == "not found"){ temp = "1.0"; }
- convert(temp, cutoff);
+ m->mothurConvert(temp, cutoff);
- temp = validParameter.validFile(parameters, "processors", false); if(temp == "not found"){ temp = "1"; }
- convert(temp, processors);
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
+ m->setProcessors(temp);
+ m->mothurConvert(temp, processors);
temp = validParameter.validFile(parameters, "compress", false); if(temp == "not found"){ temp = "F"; }
convert(temp, compress);
output = validParameter.validFile(parameters, "output", false); if(output == "not found"){ output = "column"; }
+ if (output == "phylip") { output = "lt"; }
if (((column != "") && (oldfastafile == "")) || ((column == "") && (oldfastafile != ""))) { m->mothurOut("If you provide column or oldfasta, you must provide both."); m->mothurOutEndLine(); abort=true; }
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(); }
- }
- }
- }
}
exit(1);
}
}
-
-//**********************************************************************************************************************
-
-DistanceCommand::~DistanceCommand(){
-
- for(int i=0;i<lines.size();i++){
- delete lines[i];
- }
-
-}
-
-//**********************************************************************************************************************
-
-void DistanceCommand::help(){
- try {
- m->mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
- m->mothurOut("The dist.seqs command parameters are fasta, oldfasta, column, calc, countends, output, compress, cutoff and processors. \n");
- m->mothurOut("The fasta parameter is required.\n");
- m->mothurOut("The oldfasta and column parameters allow you to append the distances calculated to the column file.\n");
- m->mothurOut("The calc parameter allows you to specify the method of calculating the distances. Your options are: nogaps, onegap or eachgap. The default is onegap.\n");
- m->mothurOut("The countends parameter allows you to specify whether to include terminal gaps in distance. Your options are: T or F. The default is T.\n");
- m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
- m->mothurOut("The output parameter allows you to specify format of your distance matrix. Options are column, lt, and square. The default is column.\n");
- m->mothurOut("The processors parameter allows you to specify number of processors to use. The default is 1.\n");
- m->mothurOut("The compress parameter allows you to indicate that you want the resulting distance file compressed. The default is false.\n");
- m->mothurOut("The dist.seqs command should be in the following format: \n");
- m->mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
- m->mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
- }
- catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "help");
- exit(1);
- }
-}
//**********************************************************************************************************************
int DistanceCommand::execute(){
try {
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
int startTime = time(NULL);
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;
-
+
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
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);
+ variables["[outputtag]"] = "phylip";
+ outputFile = getOutputFileName("phylip", variables);
+ m->mothurRemove(outputFile); outputTypes["phylip"].push_back(outputFile);
//output numSeqs to phylip formatted dist file
}else if (output == "column") { //user wants column format
- outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
+ outputFile = getOutputFileName("column", variables);
outputTypes["column"].push_back(outputFile);
//so we don't accidentally overwrite
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());
+ variables["[outputtag]"] = "square";
+ outputFile = getOutputFileName("phylip", variables);
+ m->mothurRemove(outputFile);
outputTypes["phylip"].push_back(outputFile);
}
//delete filename;
if (pid == 0) { //you are the root process
-
+
//do your part
string outputMyPart;
driverMPI(start, end, outMPI, cutoff);
- if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); return 0; }
//wait on chidren
for(int i = 1; i < processors; i++) {
- if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); return 0; }
char buf[5];
MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
//do your part
driverMPI(start, end, outMPI, cutoff);
- if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); return 0; }
char buf[5];
strcpy(buf, "done");
//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); }
- if (m->control_pressed) { outputTypes.clear(); delete distCalculator; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); return 0; }
int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
MPI_File outMPI;
//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; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); return 0; }
MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
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); }
- if (m->control_pressed) { delete distCalculator; return 0; }
+ if (m->control_pressed) { return 0; }
//tell parent you are done.
MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//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 long numDists = 0;
+
+ if (output == "square") {
+ numDists = numSeqs * numSeqs;
+ }else {
+ for(int i=0;i<numSeqs;i++){
+ for(int j=0;j<i;j++){
+ numDists++;
+ if (numDists > processors) { break; }
+ }
+ }
+ }
+
+ if (numDists < processors) { processors = numDists; }
+
for (int i = 0; i < processors; i++) {
- lines.push_back(new linePair());
+ distlinePair tempLine;
+ lines.push_back(tempLine);
if (output != "square") {
- lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
- lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
+ 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);
+ lines[i].start = int ((float(i)/float(processors)) * numSeqs);
+ lines[i].end = int ((float(i+1)/float(processors)) * numSeqs);
}
+
}
-
- createProcesses(outputFile);
-
- map<int, int>::iterator it = processIDS.begin();
- rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
- it++;
- //append and remove temp files
- for (; it != processIDS.end(); it++) {
- m->appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
- remove((outputFile + toString(it->second) + ".temp").c_str());
- }
+ 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; }
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFile); return 0; }
- delete distCalculator;
+ //set phylip file as new current phylipfile
+ string current = "";
+ itTypes = outputTypes.find("phylip");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
+ }
+
+ //set column file as new current columnfile
+ itTypes = outputTypes.find("column");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setColumnFile(current); }
+ }
m->mothurOutEndLine();
- m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();
m->mothurOut(outputFile); m->mothurOutEndLine();
m->mothurOutEndLine();
m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
/**************************************************************************************************/
void DistanceCommand::createProcesses(string filename) {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- int process = 0;
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ int process = 1;
processIDS.clear();
//loop through and create all the processes you want
int pid = fork();
if (pid > 0) {
- processIDS[lines[process]->end] = pid; //create map from line number to pid so you can append files in correct order later
+ 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){
- 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"); }
+ 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("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
+ perror(" : ");
+ for (int i=0;i<processIDS.size();i++) { int temp = processIDS[i]; kill (temp, SIGINT); }
exit(0);
- }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
+ }
}
-
+
+ //parent does its part
+ if (output != "square") { driver(lines[0].start, lines[0].end, filename, cutoff); }
+ else { driver(lines[0].start, lines[0].end, filename, "square"); }
+
+
//force parent to wait until all the processes are done
- for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
- int temp = it->second;
+ for (int i=0;i<processIDS.size();i++) {
+ 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++){
+ if (pDataArray[i]->count != (pDataArray[i]->endLine-pDataArray[i]->startLine)) {
+ m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->endLine-pDataArray[i]->startLine) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
+ }
+ 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);
+ m->mothurRemove((filename + toString(processIDS[i]) + ".temp"));
+ }
+
}
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
if (output == "lt") { outFile << endl; }
if(i % 100 == 0){
- m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");
}
}
- m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
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();
outFile << endl;
if(i % 100 == 0){
- m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");
}
}
- m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
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
}
if(i % 100 == 0){
- //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- cout << i << '\t' << (time(NULL) - startTime) << endl;
+ m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");
}
}
- //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
+ 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();
if(i % 100 == 0){
- //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- cout << i << '\t' << (time(NULL) - startTime) << endl;
- }
+ m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n"); }
//send results to parent
delete buf;
}
- //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
+
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();
if(i % 100 == 0){
- //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- cout << i << '\t' << (time(NULL) - startTime) << endl;
+ m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");
}
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);
+ m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
+ MPI_File_close(&outMPI);
+ delete distCalculator;
return 1;
}
catch(exception& e) {
string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
//use the unix sort
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
string command = "sort -n " + outputFile + " -o " + outfile;
system(command.c_str());
#else //sort using windows sort
//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;
exit(1);
}
}
-/**************************************************************************************************
+**************************************************************************************************
int DistanceCommand::convertToLowerTriangle(string outputFile) {
try{
string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
//use the unix sort
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
string command = "sort -n " + outputFile + " -o " + outfile;
system(command.c_str());
#else //sort using windows sort
//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;
exit(1);
}
}
-/**************************************************************************************************/
+**************************************************************************************************/
//its okay if the column file does not contain all the names in the fasta file, since some distance may have been above a cutoff,
//but no sequences can be in the column file that are not in oldfasta. also, if a distance is above the cutoff given then remove it.
//also check to make sure the 2 files have the same alignment length.
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;
+
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "m->appendFiles");
+ m->errorOut(e, "DistanceCommand", "sanityCheck");
exit(1);
}
}
projects / mothur.git / blobdiff