//**********************************************************************************************************************
vector<string> DistanceCommand::setParameters(){
try {
- CommandParameter pcolumn("column", "InputTypes", "", "", "none", "none", "OldFastaColumn",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",false,true); parameters.push_back(pfasta);
- CommandParameter poutput("output", "Multiple", "column-lt-square", "column", "", "", "",false,false); 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); parameters.push_back(pprocessors);
- CommandParameter pcutoff("cutoff", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pcutoff);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ 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); }
}
}
//**********************************************************************************************************************
+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);
+ }
+}
+//**********************************************************************************************************************
DistanceCommand::DistanceCommand(){
try {
abort = true; calledHelp = true;
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 = m->getProcessors(); }
m->setProcessors(temp);
- convert(temp, processors);
+ 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 (!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";
+ 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
m->mothurRemove(outputFile);
}else { //assume square
- outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
+ variables["[outputtag]"] = "square";
+ outputFile = getOutputFileName("phylip", variables);
m->mothurRemove(outputFile);
outputTypes["phylip"].push_back(outputFile);
}
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 int numDists = 0;
+ unsigned long long numDists = 0;
if (output == "square") {
numDists = numSeqs * numSeqs;
}
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)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
int process = 1;
processIDS.clear();
//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];
}
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;
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;
}
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;
}
}
/**************************************************************************************************/
/////// 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(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;
}
/**************************************************************************************************/
/////// 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(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;
+ m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
+
MPI_File_close(&outMPI);
delete distCalculator;
return 1;
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
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
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.
projects / mothur.git / blobdiff