CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
CommandParameter psubsample("subsample", "String", "", "", "", "", "","phylip",false,false); parameters.push_back(psubsample);
CommandParameter pdistance("distance", "Boolean", "", "F", "", "", "","phylip",false,false); parameters.push_back(pdistance);
- CommandParameter pcalc("calc", "Multiple", "sharedchao-sharedsobs-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-whittaker-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-structchi2-hamming-gower-memchi2-memchord-memeuclidean-mempearson", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan", "", "", "","",true,false,true); parameters.push_back(pcalc);
+ CommandParameter pcalc("calc", "Multiple", "sharedchao-sharedsobs-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-whittaker-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-structchi2-hamming-gower-memchi2-memchord-memeuclidean-mempearson-jsd-rjsd", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan", "", "", "","",true,false,true); parameters.push_back(pcalc);
CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "","",false,false); parameters.push_back(poutput);
CommandParameter pall("all", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pall);
CommandParameter piters("iters", "Number", "", "1000", "", "", "","",false,false); parameters.push_back(piters);
string temp = validParameter.validFile(parameters, "all", false); if (temp == "not found") { temp = "false"; }
all = m->isTrue(temp);
- temp = validParameter.validFile(parameters, "distance", false); if (temp == "not found") { temp = "false"; }
- createPhylip = m->isTrue(temp);
-
temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "1000"; }
m->mothurConvert(temp, iters);
- output = validParameter.validFile(parameters, "output", false); if(output == "not found"){ output = "lt"; }
+ output = validParameter.validFile(parameters, "output", false);
+ if(output == "not found"){ output = "lt"; }
+ else { createPhylip = true; }
if ((output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are lt and square. I will use lt."); m->mothurOutEndLine(); output = "lt"; }
temp = validParameter.validFile(parameters, "subsample", false); if (temp == "not found") { temp = "F"; }
if (subsample == false) { iters = 0; }
+ temp = validParameter.validFile(parameters, "distance", false); if (temp == "not found") { temp = "false"; }
+ createPhylip = m->isTrue(temp);
+ if (subsample) { createPhylip = true; }
+
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
m->setProcessors(temp);
m->mothurConvert(temp, processors);
sumCalculators.push_back(new MemEuclidean());
}else if (Estimators[i] == "mempearson") {
sumCalculators.push_back(new MemPearson());
+ }else if (Estimators[i] == "jsd") {
+ sumCalculators.push_back(new JSD());
+ }else if (Estimators[i] == "rjsd") {
+ sumCalculators.push_back(new RJSD());
}
}
}
if(processors == 1){
- driver(thisLookup, 0, numGroups, sumFileName+".temp", sumAllFileName+".temp", calcDists);
+ driver(thisItersLookup, 0, numGroups, sumFileName+".temp", sumAllFileName+".temp", calcDists);
m->appendFiles((sumFileName + ".temp"), sumFileName);
m->mothurRemove((sumFileName + ".temp"));
if (mult) {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//loop through and create all the processes you want
while (process != processors) {
- int pid = fork();
+ pid_t pid = fork();
if (pid > 0) {
processIDS.push_back(pid);
process++;
}else if (pid == 0){
- driver(thisLookup, lines[process].start, lines[process].end, sumFileName + toString(getpid()) + ".temp", sumAllFileName + toString(getpid()) + ".temp", calcDists);
+ driver(thisItersLookup, lines[process].start, lines[process].end, sumFileName + m->mothurGetpid(process) + ".temp", sumAllFileName + m->mothurGetpid(process) + ".temp", calcDists);
//only do this if you want a distance file
if (createPhylip) {
- string tempdistFileName = m->getRootName(m->getSimpleName(sumFileName)) + toString(getpid()) + ".dist";
+ string tempdistFileName = m->getRootName(m->getSimpleName(sumFileName)) + m->mothurGetpid(process) + ".dist";
ofstream outtemp;
m->openOutputFile(tempdistFileName, outtemp);
}
//parent do your part
- driver(thisLookup, lines[0].start, lines[0].end, sumFileName + toString(getpid()) + ".temp", sumAllFileName + toString(getpid()) + ".temp", calcDists);
- m->appendFiles((sumFileName + toString(getpid()) + ".temp"), sumFileName);
- m->mothurRemove((sumFileName + toString(getpid()) + ".temp"));
- if (mult) { m->appendFiles((sumAllFileName + toString(getpid()) + ".temp"), sumAllFileName); }
+ driver(thisItersLookup, lines[0].start, lines[0].end, sumFileName + m->mothurGetpid(process) + ".temp", sumAllFileName + m->mothurGetpid(process) + ".temp", calcDists);
+ m->appendFiles((sumFileName + m->mothurGetpid(process) + ".temp"), sumFileName);
+ m->mothurRemove((sumFileName + m->mothurGetpid(process) + ".temp"));
+ if (mult) { m->appendFiles((sumAllFileName + m->mothurGetpid(process) + ".temp"), sumAllFileName); }
//force parent to wait until all the processes are done
for (int i = 0; i < processIDS.size(); i++) {
vector<summarySharedData*> pDataArray;
DWORD dwThreadIdArray[processors-1];
- HANDLE hThreadArray[processors-1];
+ HANDLE hThreadArray[processors-1];
//Create processor worker threads.
for( int i=1; i<processors; i++ ){
temp->setGroup(thisLookup[k]->getGroup());
newLookup.push_back(temp);
}
+
//for each bin
- for (int k = 0; k < thisLookup[0]->getNumBins(); k++) {
+ for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) { delete newLookup[j]; } return 0; }
- for (int j = 0; j < thisLookup.size(); j++) { newLookup[j]->push_back(thisLookup[j]->getAbundance(k), thisLookup[j]->getGroup()); }
+ for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
}
// Allocate memory for thread data.
}
//parent do your part
- driver(thisLookup, lines[0].start, lines[0].end, sumFileName +"0.temp", sumAllFileName + "0.temp", calcDists);
+ driver(thisItersLookup, lines[0].start, lines[0].end, sumFileName +"0.temp", sumAllFileName + "0.temp", calcDists);
m->appendFiles((sumFileName + "0.temp"), sumFileName);
m->mothurRemove((sumFileName + "0.temp"));
if (mult) { m->appendFiles((sumAllFileName + "0.temp"), sumAllFileName); }
//Close all thread handles and free memory allocations.
for(int i=0; i < pDataArray.size(); i++){
+ if (pDataArray[i]->count != (pDataArray[i]->end-pDataArray[i]->start)) {
+ m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end-pDataArray[i]->start) + " groups assigned to it, quitting. \n"); m->control_pressed = true;
+ }
m->appendFiles((sumFileName + toString(processIDS[i]) + ".temp"), sumFileName);
m->mothurRemove((sumFileName + toString(processIDS[i]) + ".temp"));
- for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) { delete pDataArray[i]->thisLookup[j]; }
+ for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) { delete pDataArray[i]->thisLookup[j]; }
if (createPhylip) {
for (int k = 0; k < calcDists.size(); k++) {
variables["[calc]"] = sumCalculators[i]->getName();
variables["[distance]"] = thisLookup[0]->getLabel();
variables["[outputtag]"] = output;
+ variables["[tag2]"] = "";
string distFileName = getOutputFileName("phylip",variables);
outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
ofstream outDist;
for (int i = 0; i < calcDists.size(); i++) { calcDists[i].clear(); }
}
- if (iters != 1) {
+ if (iters != 0) {
//we need to find the average distance and standard deviation for each groups distance
-
- vector< vector<seqDist> > calcAverages; calcAverages.resize(sumCalculators.size());
- for (int i = 0; i < calcAverages.size(); i++) { //initialize sums to zero.
- calcAverages[i].resize(calcDistsTotals[0][i].size());
-
- for (int j = 0; j < calcAverages[i].size(); j++) {
- calcAverages[i][j].seq1 = calcDists[i][j].seq1;
- calcAverages[i][j].seq2 = calcDists[i][j].seq2;
- calcAverages[i][j].dist = 0.0;
- }
- }
-
- for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
- for (int i = 0; i < calcAverages.size(); i++) { //initialize sums to zero.
- for (int j = 0; j < calcAverages[i].size(); j++) {
- calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist;
- }
- }
- }
-
- for (int i = 0; i < calcAverages.size(); i++) { //finds average.
- for (int j = 0; j < calcAverages[i].size(); j++) {
- calcAverages[i][j].dist /= (float) iters;
- }
- }
+ vector< vector<seqDist> > calcAverages = m->getAverages(calcDistsTotals);
//find standard deviation
- vector< vector<seqDist> > stdDev; stdDev.resize(sumCalculators.size());
- for (int i = 0; i < stdDev.size(); i++) { //initialize sums to zero.
- stdDev[i].resize(calcDistsTotals[0][i].size());
-
- for (int j = 0; j < stdDev[i].size(); j++) {
- stdDev[i][j].seq1 = calcDists[i][j].seq1;
- stdDev[i][j].seq2 = calcDists[i][j].seq2;
- stdDev[i][j].dist = 0.0;
- }
- }
-
- for (int thisIter = 0; thisIter < iters; thisIter++) { //compute the difference of each dist from the mean, and square the result of each
- for (int i = 0; i < stdDev.size(); i++) {
- for (int j = 0; j < stdDev[i].size(); j++) {
- stdDev[i][j].dist += ((calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist) * (calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist));
- }
- }
- }
-
- for (int i = 0; i < stdDev.size(); i++) { //finds average.
- for (int j = 0; j < stdDev[i].size(); j++) {
- stdDev[i][j].dist /= (float) iters;
- stdDev[i][j].dist = sqrt(stdDev[i][j].dist);
- }
- }
+ vector< vector<seqDist> > stdDev = m->getStandardDeviation(calcDistsTotals, calcAverages);
//print results
for (int i = 0; i < calcDists.size(); i++) {
projects / mothur.git / blobdiff