#include "pintail.h"
#include "ignoregaps.h"
+#include "eachgapdist.h"
+//********************************************************************************************************************
+//sorts lowest to highest
+inline bool compareQuanMembers(quanMember left, quanMember right){
+ return (left.score < right.score);
+}
//***************************************************************************************************************
-Pintail::Pintail(string name) {
+Pintail::Pintail(string filename, string temp, bool f, int p, string mask, string cons, string q, int win, int inc, string o) : Chimera() {
try {
- fastafile = name;
+
+ fastafile = filename;
+ templateFileName = temp; templateSeqs = readSeqs(temp);
+ filter = f;
+ processors = p;
+ setMask(mask);
+ consfile = cons;
+ quanfile = q;
+ window = win;
+ increment = inc;
+ outputDir = o;
+
+ distcalculator = new eachGapDist();
+ decalc = new DeCalculator();
+
+ doPrep();
}
catch(exception& e) {
- errorOut(e, "Pintail", "Pintail");
+ m->errorOut(e, "Pintail", "Pintail");
exit(1);
}
+
}
//***************************************************************************************************************
Pintail::~Pintail() {
- try {
- for (int i = 0; i < querySeqs.size(); i++) { delete querySeqs[i]; }
- for (int i = 0; i < templateSeqs.size(); i++) { delete templateSeqs[i]; }
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "~Pintail");
- exit(1);
- }
-}
-//***************************************************************************************************************
-void Pintail::print(ostream& out) {
try {
- for (itCoef = DE.begin(); itCoef != DE.end(); itCoef++) {
-
- out << itCoef->first->getName() << '\t' << itCoef->second << endl;
- out << "Observed\t";
-
- itObsDist = obsDistance.find(itCoef->first);
- for (int i = 0; i < itObsDist->second.size(); i++) { out << itObsDist->second[i] << '\t'; }
- out << endl;
-
- out << "Expected\t";
-
- itExpDist = expectedDistance.find(itCoef->first);
- for (int i = 0; i < itExpDist->second.size(); i++) { out << itExpDist->second[i] << '\t'; }
- out << endl;
-
- }
-
-
+ delete distcalculator;
+ delete decalc;
}
catch(exception& e) {
- errorOut(e, "Pintail", "print");
+ m->errorOut(e, "Pintail", "~Pintail");
exit(1);
}
}
-
//***************************************************************************************************************
-void Pintail::getChimeras() {
+int Pintail::doPrep() {
try {
- distCalculator = new ignoreGaps();
-
- //read in query sequences and subject sequences
- mothurOut("Reading sequences and template file... "); cout.flush();
- querySeqs = readSeqs(fastafile);
- templateSeqs = readSeqs(templateFile);
- mothurOut("Done."); mothurOutEndLine();
-
- int numSeqs = querySeqs.size();
-
- //if window is set to default
- if (window == 0) { if (querySeqs[0]->getAligned().length() > 800) { setWindow(200); }
- else{ setWindow((querySeqs[0]->getAligned().length() / 4)); } }
- else if (window > (querySeqs[0]->getAligned().length() / 4)) {
- mothurOut("You have selected to large a window size for you sequences. I will choose a smaller window."); mothurOutEndLine();
- setWindow((querySeqs[0]->getAligned().length() / 4));
- }
-
- //calculate number of iters
- iters = (querySeqs[0]->getAligned().length() - window + 1) / increment;
-cout << "length = " << querySeqs[0]->getAligned().length() << " window = " << window << " increment = " << increment << " iters = " << iters << endl;
- int linesPerProcess = processors / numSeqs;
-
- //find breakup of sequences for all times we will Parallelize
- if (processors == 1) { lines.push_back(new linePair(0, numSeqs)); }
- else {
- //fill line pairs
- for (int i = 0; i < (processors-1); i++) {
- lines.push_back(new linePair((i*linesPerProcess), ((i*linesPerProcess) + linesPerProcess)));
+ mergedFilterString = "";
+ windowSizesTemplate.resize(templateSeqs.size(), window);
+ quantiles.resize(100); //one for every percent mismatch
+ quantilesMembers.resize(100); //one for every percent mismatch
+
+ //if the user does not enter a mask then you want to keep all the spots in the alignment
+ if (seqMask.length() == 0) { decalc->setAlignmentLength(templateSeqs[0]->getAligned().length()); }
+ else { decalc->setAlignmentLength(seqMask.length()); }
+
+ decalc->setMask(seqMask);
+
+ #ifdef USE_MPI
+ //do nothing
+ #else
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ //find breakup of templatefile for quantiles
+ if (processors == 1) { templateLines.push_back(new linePair(0, templateSeqs.size())); }
+ else {
+ for (int i = 0; i < processors; i++) {
+ templateLines.push_back(new linePair());
+ templateLines[i]->start = int (sqrt(float(i)/float(processors)) * templateSeqs.size());
+ templateLines[i]->end = int (sqrt(float(i+1)/float(processors)) * templateSeqs.size());
+ }
}
- //this is necessary to get remainder of processors / numSeqs so you don't miss any lines at the end
- int i = processors - 1;
- lines.push_back(new linePair((i*linesPerProcess), numSeqs));
- }
-
- //map query sequences to their most similiar sequences in the template - Parallelized
- mothurOut("Finding closest sequence in template to each sequence... "); cout.flush();
- if (processors == 1) { findPairs(lines[0]->start, lines[0]->end); }
- else { createProcessesPairs(); }
- mothurOut("Done."); mothurOutEndLine();
-
- //find Oqs for each sequence - the observed distance in each window - Parallelized
- mothurOut("Calculating observed percentage differences for each sequence... "); cout.flush();
- if (processors == 1) { calcObserved(lines[0]->start, lines[0]->end); }
- else { createProcessesObserved(); }
- mothurOut("Done."); mothurOutEndLine();
-
- //find P
- mothurOut("Calculating expected percentage differences for each sequence... "); cout.flush();
- vector<float> probabilityProfile = calcFreq(templateSeqs);
+ #else
+ templateLines.push_back(new linePair(0, templateSeqs.size()));
+ #endif
+ #endif
+
+ m->mothurOut("Getting conservation... "); cout.flush();
+ if (consfile == "") {
+ m->mothurOut("Calculating probability of conservation for your template sequences. This can take a while... I will output the frequency of the highest base in each position to a .freq file so that you can input them using the conservation parameter next time you run this command. Providing the .freq file will improve speed. "); cout.flush();
+ probabilityProfile = decalc->calcFreq(templateSeqs, templateFileName);
+ if (m->control_pressed) { return 0; }
+ m->mothurOut("Done."); m->mothurOutEndLine();
+ }else { probabilityProfile = readFreq(); m->mothurOut("Done."); }
+ m->mothurOutEndLine();
//make P into Q
- for (int i = 0; i < probabilityProfile.size(); i++) { probabilityProfile[i] = 1 - probabilityProfile[i]; }
-
- //find Qav
- averageProbability = findQav(probabilityProfile);
-
- //find Coefficient - maps a sequence to its coefficient
- seqCoef = getCoef(averageProbability);
-
- //find Eqs for each sequence - the expected distance in each window - Parallelized
- if (processors == 1) { calcExpected(lines[0]->start, lines[0]->end); }
- else { createProcessesExpected(); }
- mothurOut("Done."); mothurOutEndLine();
-
- //find deviation - Parallelized
- mothurOut("Finding deviation from expected... "); cout.flush();
- if (processors == 1) { calcDE(lines[0]->start, lines[0]->end); }
- else { createProcessesDE(); }
- mothurOut("Done."); mothurOutEndLine();
+ for (int i = 0; i < probabilityProfile.size(); i++) { probabilityProfile[i] = 1 - probabilityProfile[i]; } //
+ bool reRead = false;
+ //create filter if needed for later
+ if (filter) {
+
+ //read in all query seqs
+ vector<Sequence*> tempQuerySeqs = readSeqs(fastafile);
- //free memory
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; }
- delete distCalculator;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "getChimeras");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-
-vector<Sequence*> Pintail::readSeqs(string file) {
- try {
+ vector<Sequence*> temp;
+ //merge query seqs and template seqs
+ temp = templateSeqs;
+ for (int i = 0; i < tempQuerySeqs.size(); i++) { temp.push_back(tempQuerySeqs[i]); }
- ifstream in;
- openInputFile(file, in);
- vector<Sequence*> container;
-
- //read in seqs and store in vector
- while(!in.eof()){
- Sequence* current = new Sequence(in);
-
- if (current->getAligned() == "") { current->setAligned(current->getUnaligned()); }
- //takes out stuff is needed
- current->setUnaligned(current->getUnaligned());
+ if (seqMask != "") {
+ reRead = true;
+ //mask templates
+ for (int i = 0; i < temp.size(); i++) {
+ if (m->control_pressed) {
+ for (int i = 0; i < tempQuerySeqs.size(); i++) { delete tempQuerySeqs[i]; }
+ return 0;
+ }
+ decalc->runMask(temp[i]);
+ }
+ }
+
+ mergedFilterString = createFilter(temp, 0.5);
- container.push_back(current);
+ if (m->control_pressed) {
+ for (int i = 0; i < tempQuerySeqs.size(); i++) { delete tempQuerySeqs[i]; }
+ return 0;
+ }
- gobble(in);
+ //reread template seqs
+ for (int i = 0; i < tempQuerySeqs.size(); i++) { delete tempQuerySeqs[i]; }
}
- in.close();
- return container;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "readSeqs");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-//calculate the distances from each query sequence to all sequences in the template to find the closest sequence
-void Pintail::findPairs(int start, int end) {
- try {
-
- for(int i = start; i < end; i++){
+ //quantiles are used to determine whether the de values found indicate a chimera
+ //if you have to calculate them, its time intensive because you are finding the de and deviation values for each
+ //combination of sequences in the template
+ if (quanfile != "") {
+ quantiles = readQuantiles();
+ }else {
+ if ((!filter) && (seqMask != "")) { //if you didn't filter but you want to mask. if you filtered then you did mask first above.
+ reRead = true;
+ //mask templates
+ for (int i = 0; i < templateSeqs.size(); i++) {
+ if (m->control_pressed) { return 0; }
+ decalc->runMask(templateSeqs[i]);
+ }
+ }
+
+ if (filter) {
+ reRead = true;
+ for (int i = 0; i < templateSeqs.size(); i++) {
+ if (m->control_pressed) { return 0; }
+ runFilter(templateSeqs[i]);
+ }
+ }
+
+ m->mothurOut("Calculating quantiles for your template. This can take a while... I will output the quantiles to a .quan file that you can input them using the quantiles parameter next time you run this command. Providing the .quan file will dramatically improve speed. "); cout.flush();
+ if (processors == 1) {
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, 0, templateSeqs.size());
+ }else { createProcessesQuan(); }
- float smallest = 10000.0;
- Sequence query = *(querySeqs[i]);
+ if (m->control_pressed) { return 0; }
+
+ string noOutliers, outliers;
+
+ if ((!filter) && (seqMask == "")) {
+ noOutliers = m->getRootName(m->getSimpleName(templateFileName)) + "pintail.quan";
+ }else if ((!filter) && (seqMask != "")) {
+ noOutliers =m->getRootName(m->getSimpleName(templateFileName)) + "pintail.masked.quan";
+ }else if ((filter) && (seqMask != "")) {
+ noOutliers = m->getRootName(m->getSimpleName(templateFileName)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastafile)) + "masked.quan";
+ }else if ((filter) && (seqMask == "")) {
+ noOutliers = m->getRootName(m->getSimpleName(templateFileName)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastafile)) + "quan";
+ }
+
+ decalc->removeObviousOutliers(quantilesMembers, templateSeqs.size());
+
+ if (m->control_pressed) { return 0; }
- for(int j = 0; j < templateSeqs.size(); j++){
+ string outputString = "#" + m->getVersion() + "\n";
+
+ //adjust quantiles
+ for (int i = 0; i < quantilesMembers.size(); i++) {
+ vector<float> temp;
- Sequence temp = *(templateSeqs[j]);
+ if (quantilesMembers[i].size() == 0) {
+ //in case this is not a distance found in your template files
+ for (int g = 0; g < 6; g++) {
+ temp.push_back(0.0);
+ }
+ }else{
+
+ sort(quantilesMembers[i].begin(), quantilesMembers[i].end());
+
+ //save 10%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.10)]);
+ //save 25%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.25)]);
+ //save 50%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.5)]);
+ //save 75%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.75)]);
+ //save 95%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.95)]);
+ //save 99%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.99)]);
+
+ }
- distCalculator->calcDist(query, temp);
- float dist = distCalculator->getDist();
+ //output quan value
+ outputString += toString(i+1) + "\t";
+ for (int u = 0; u < temp.size(); u++) { outputString += toString(temp[u]) + "\t"; }
+ outputString += "\n";
- if (dist < smallest) {
+ quantiles[i] = temp;
- bestfit[querySeqs[i]] = templateSeqs[j];
- smallest = dist;
- }
}
+
+ printQuanFile(noOutliers, outputString);
+
+ //free memory
+ quantilesMembers.clear();
+
+ m->mothurOut("Done."); m->mothurOutEndLine();
}
+ if (reRead) {
+ for (int i = 0; i < templateSeqs.size(); i++) { delete templateSeqs[i]; }
+ templateSeqs.clear();
+ templateSeqs = readSeqs(templateFileName);
+ }
+
+
+ //free memory
+ for (int i = 0; i < templateLines.size(); i++) { delete templateLines[i]; }
+
+ return 0;
+
}
catch(exception& e) {
- errorOut(e, "Pintail", "findPairs");
+ m->errorOut(e, "Pintail", "doPrep");
exit(1);
}
}
//***************************************************************************************************************
-void Pintail::calcObserved(int start, int end) {
+Sequence Pintail::print(ostream& out, ostream& outAcc) {
try {
-
-
- for(int i = start; i < end; i++){
- itBest = bestfit.find(querySeqs[i]);
- Sequence* query;
- Sequence* subject;
+ int index = ceil(deviation);
- if (itBest != bestfit.end()) {
- query = itBest->first;
- subject = itBest->second;
- }else{ mothurOut("Error in calcObserved"); mothurOutEndLine(); }
-//cout << query->getName() << '\t' << subject->getName() << endl;
-
- int startpoint = 0;
- for (int m = 0; m < iters; m++) {
-
- string seqFrag = query->getAligned().substr(startpoint, window);
- string seqFragsub = subject->getAligned().substr(startpoint, window);
-
- int diff = 0;
- for (int b = 0; b < seqFrag.length(); b++) {
-
- //if this is not a gap
- if ((isalpha(seqFrag[b])) && (isalpha(seqFragsub[b]))) {
- //and they are different - penalize
- if (seqFrag[b] != seqFragsub[b]) { diff++; }
- }
- }
-
- //percentage of mismatched bases
- float dist = diff / (float)seqFrag.length();
-
- obsDistance[query].push_back(dist);
-
- startpoint += increment;
+ //is your DE value higher than the 95%
+ string chimera;
+ if (index != 0) { //if index is 0 then its an exact match to a template seq
+ if (quantiles[index][4] == 0.0) {
+ chimera = "Your template does not include sequences that provide quantile values at distance " + toString(index);
+ }else {
+ if (DE > quantiles[index][4]) { chimera = "Yes"; }
+ else { chimera = "No"; }
}
+ }else{ chimera = "No"; }
+
+ out << querySeq->getName() << '\t' << "div: " << deviation << "\tstDev: " << DE << "\tchimera flag: " << chimera << endl;
+ if (chimera == "Yes") {
+ m->mothurOut(querySeq->getName() + "\tdiv: " + toString(deviation) + "\tstDev: " + toString(DE) + "\tchimera flag: " + chimera); m->mothurOutEndLine();
+ outAcc << querySeq->getName() << endl;
}
+ out << "Observed\t";
+
+ for (int j = 0; j < obsDistance.size(); j++) { out << obsDistance[j] << '\t'; }
+ out << endl;
+
+ out << "Expected\t";
+
+ for (int m = 0; m < expectedDistance.size(); m++) { out << expectedDistance[m] << '\t'; }
+ out << endl;
+
+ return *querySeq;
}
catch(exception& e) {
- errorOut(e, "Pintail", "calcObserved");
+ m->errorOut(e, "Pintail", "print");
exit(1);
}
}
-
+#ifdef USE_MPI
//***************************************************************************************************************
-void Pintail::calcExpected(int start, int end) {
+Sequence Pintail::print(MPI_File& out, MPI_File& outAcc) {
try {
-
- //for each sequence
- for(int i = start; i < end; i++){
-
- itCoef = seqCoef.find(querySeqs[i]);
- float coef = itCoef->second;
-
- //for each window
- vector<float> queryExpected;
- for (int m = 0; m < iters; m++) {
- float expected = averageProbability[m] * coef;
- queryExpected.push_back(expected);
-//cout << "average variabilty over window = " << averageProbability[m] << " coef = " << coef << " ei = " << expected << '\t' << "window = " << m << endl;
+ string outputString = "";
+ int index = ceil(deviation);
+
+ //is your DE value higher than the 95%
+ string chimera;
+ if (index != 0) { //if index is 0 then its an exact match to a template seq
+ if (quantiles[index][4] == 0.0) {
+ chimera = "Your template does not include sequences that provide quantile values at distance " + toString(index);
+ }else {
+ if (DE > quantiles[index][4]) { chimera = "Yes"; }
+ else { chimera = "No"; }
}
+ }else{ chimera = "No"; }
+
+ outputString += querySeq->getName() + "\tdiv: " + toString(deviation) + "\tstDev: " + toString(DE) + "\tchimera flag: " + chimera + "\n";
+ if (chimera == "Yes") {
+ cout << querySeq->getName() << "\tdiv: " << toString(deviation) << "\tstDev: " << toString(DE) << "\tchimera flag: " << chimera << endl;
+ string outAccString = querySeq->getName() + "\n";
- expectedDistance[querySeqs[i]] = queryExpected;
-
+ MPI_Status statusAcc;
+ int length = outAccString.length();
+ char* buf = new char[length];
+ memcpy(buf, outAccString.c_str(), length);
+
+ MPI_File_write_shared(outAcc, buf, length, MPI_CHAR, &statusAcc);
+ delete buf;
+
+ return *querySeq;
}
+ outputString += "Observed\t";
+
+ for (int j = 0; j < obsDistance.size(); j++) { outputString += toString(obsDistance[j]) + "\t"; }
+ outputString += "\n";
+
+ outputString += "Expected\t";
+
+ for (int m = 0; m < expectedDistance.size(); m++) { outputString += toString(expectedDistance[m]) + "\t"; }
+ outputString += "\n";
+
+ MPI_Status status;
+ int length = outputString.length();
+ char* buf2 = new char[length];
+ memcpy(buf2, outputString.c_str(), length);
+ MPI_File_write_shared(out, buf2, length, MPI_CHAR, &status);
+ delete buf2;
+
+ return *querySeq;
}
catch(exception& e) {
- errorOut(e, "Pintail", "calcExpected");
+ m->errorOut(e, "Pintail", "print");
exit(1);
}
}
+#endif
//***************************************************************************************************************
-void Pintail::calcDE(int start, int end) {
+int Pintail::getChimeras(Sequence* query) {
try {
-
-
- //for each sequence
- for(int i = start; i < end; i++){
-
- itObsDist = obsDistance.find(querySeqs[i]);
- vector<float> obs = itObsDist->second;
-
- itExpDist = expectedDistance.find(querySeqs[i]);
- vector<float> exp = itExpDist->second;
-// cout << "difference between obs and exp = " << abs(obs[m] - exp[m]) << endl;
- //for each window
- float sum = 0.0; //sum = sum from 1 to m of (oi-ei)^2
- for (int m = 0; m < iters; m++) { sum += ((obs[m] - exp[m]) * (obs[m] - exp[m])); }
-
- float de = sqrt((sum / (iters - 1)));
-
- DE[querySeqs[i]] = de;
+ querySeq = query;
+ trimmed.clear();
+ windowSizes = window;
+
+ //find pairs has to be done before a mask
+ bestfit = findPairs(query);
+
+ if (m->control_pressed) { return 0; }
+
+ //if they mask
+ if (seqMask != "") {
+ decalc->runMask(query);
+ decalc->runMask(bestfit);
+ }
+
+ if (filter) { //must be done after a mask
+ runFilter(query);
+ runFilter(bestfit);
}
+
+
+ //trim seq
+ decalc->trimSeqs(query, bestfit, trimmed);
+
+ //find windows
+ it = trimmed.begin();
+ windowsForeachQuery = decalc->findWindows(query, it->first, it->second, windowSizes, increment);
+
+ //find observed distance
+ obsDistance = decalc->calcObserved(query, bestfit, windowsForeachQuery, windowSizes);
+
+ if (m->control_pressed) { return 0; }
+ Qav = decalc->findQav(windowsForeachQuery, windowSizes, probabilityProfile);
+
+ if (m->control_pressed) { return 0; }
+
+ //find alpha
+ seqCoef = decalc->getCoef(obsDistance, Qav);
+
+ //calculating expected distance
+ expectedDistance = decalc->calcExpected(Qav, seqCoef);
+
+ if (m->control_pressed) { return 0; }
+
+ //finding de
+ DE = decalc->calcDE(obsDistance, expectedDistance);
+
+ if (m->control_pressed) { return 0; }
+
+ //find distance between query and closest match
+ it = trimmed.begin();
+ deviation = decalc->calcDist(query, bestfit, it->first, it->second);
+
+ delete bestfit;
+
+ return 0;
}
catch(exception& e) {
- errorOut(e, "Pintail", "calcDE");
+ m->errorOut(e, "Pintail", "getChimeras");
exit(1);
}
}
//***************************************************************************************************************
-vector<float> Pintail::calcFreq(vector<Sequence*> seqs) {
+vector<float> Pintail::readFreq() {
try {
+ //read in probabilities and store in vector
+ int pos; float num;
vector<float> prob;
+ set<int> h = decalc->getPos(); //positions of bases in masking sequence
- //at each position in the sequence
- for (int i = 0; i < seqs[0]->getAligned().length(); i++) {
+ #ifdef USE_MPI
- vector<int> freq; freq.resize(4,0);
- int gaps = 0;
-
- //find the frequency of each nucleotide
- for (int j = 0; j < seqs.size(); j++) {
-
- char value = seqs[j]->getAligned()[i];
+ MPI_File inMPI;
+ MPI_Offset size;
+ MPI_Status status;
+
+ //char* inFileName = new char[consfile.length()];
+ //memcpy(inFileName, consfile.c_str(), consfile.length());
+
+ char inFileName[1024];
+ strcpy(inFileName, consfile.c_str());
+
+ MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
+ MPI_File_get_size(inMPI, &size);
+ //delete inFileName;
+
+ char* buffer = new char[size];
+ MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
+
+ string tempBuf = buffer;
+ delete buffer;
+
+ if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size); }
+ istringstream iss (tempBuf,istringstream::in);
+
+ //read version
+ string line = m->getline(iss); m->gobble(iss);
+
+ while(!iss.eof()) {
+ iss >> pos >> num;
+
+ if (h.count(pos) > 0) {
+ float Pi;
+ Pi = (num - 0.25) / 0.75;
- if(toupper(value) == 'A') { freq[0]++; }
- else if(toupper(value) == 'T' || toupper(value) == 'U') { freq[1]++; }
- else if(toupper(value) == 'G') { freq[2]++; }
- else if(toupper(value) == 'C') { freq[3]++; }
- else { gaps++; }
+ //cannot have probability less than 0.
+ if (Pi < 0) { Pi = 0.0; }
+
+ //do you want this spot
+ prob.push_back(Pi);
}
- //find base with highest frequency
- int highest = 0;
- for (int m = 0; m < freq.size(); m++) { if (freq[m] > highest) { highest = freq[m]; } }
-
- //add in gaps - so you can effectively "ignore them"
- highest += gaps;
-
- float highFreq = highest / (float) seqs.size();
-
- float Pi;
- Pi = (highFreq - 0.25) / 0.75;
-
- prob.push_back(Pi);
+ m->gobble(iss);
}
+
+ MPI_File_close(&inMPI);
- return prob;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcFreq");
- exit(1);
- }
-}
-//***************************************************************************************************************
-vector<float> Pintail::findQav(vector<float> prob) {
- try {
- vector<float> averages;
+ #else
+
+ ifstream in;
+ m->openInputFile(consfile, in);
- //for each window find average
- int startpoint = 0;
- for (int m = 0; m < iters; m++) {
+ //read version
+ string line = m->getline(in); m->gobble(in);
+
+ while(!in.eof()){
- float average = 0.0;
- for (int i = startpoint; i < (startpoint+window); i++) { average += prob[i]; }
+ in >> pos >> num;
- average = average / window;
-//cout << average << endl;
- //save this windows average
- averages.push_back(average);
-
- startpoint += increment;
+ if (h.count(pos) > 0) {
+ float Pi;
+ Pi = (num - 0.25) / 0.75;
+
+ //cannot have probability less than 0.
+ if (Pi < 0) { Pi = 0.0; }
+
+ //do you want this spot
+ prob.push_back(Pi);
+ }
+
+ m->gobble(in);
}
+ in.close();
+
+ #endif
+
+ return prob;
- return averages;
}
catch(exception& e) {
- errorOut(e, "Pintail", "findQav");
+ m->errorOut(e, "Pintail", "readFreq");
exit(1);
}
}
+
//***************************************************************************************************************
-map<Sequence*, float> Pintail::getCoef(vector<float> prob) {
+//calculate the distances from each query sequence to all sequences in the template to find the closest sequence
+Sequence* Pintail::findPairs(Sequence* q) {
try {
- map<Sequence*, float> coefs;
-
- //find average prob
- float probAverage = 0.0;
- for (int i = 0; i < prob.size(); i++) { probAverage += prob[i]; }
- probAverage = probAverage / (float) prob.size();
-cout << "(sum of ai) / m = " << probAverage << endl;
- //find a coef for each sequence
- map<Sequence*, vector<float> >::iterator it;
- for (it = obsDistance.begin(); it != obsDistance.end(); it++) {
-
- vector<float> temp = it->second;
- Sequence* tempSeq = it->first;
-
- //find observed average
- float obsAverage = 0.0;
- for (int i = 0; i < temp.size(); i++) { obsAverage += temp[i]; }
- obsAverage = obsAverage / (float) temp.size();
-cout << tempSeq->getName() << '\t' << obsAverage << endl;
- float coef = obsAverage / probAverage;
-cout << tempSeq->getName() << '\t' << "coef = " << coef << endl;
- //save this sequences coefficient
- coefs[tempSeq] = coef;
- }
-
- return coefs;
+ Sequence* seqsMatches;
+
+ seqsMatches = decalc->findClosest(q, templateSeqs);
+ return seqsMatches;
+
}
catch(exception& e) {
- errorOut(e, "Pintail", "getCoef");
+ m->errorOut(e, "Pintail", "findPairs");
exit(1);
}
}
-
-/**************************************************************************************************/
-
-void Pintail::createProcessesPairs() {
+//**************************************************************************************************
+void Pintail::createProcessesQuan() {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- int process = 0;
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ int process = 1;
vector<int> processIDS;
-
+
//loop through and create all the processes you want
while (process != processors) {
int pid = fork();
processIDS.push_back(pid);
process++;
}else if (pid == 0){
- findPairs(lines[process]->start, lines[process]->end);
+
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, templateLines[process]->start, templateLines[process]->end);
+
+ //write out data to file so parent can read it
+ ofstream out;
+ string s = toString(getpid()) + ".temp";
+ m->openOutputFile(s, out);
+
+ //output observed distances
+ for (int i = 0; i < quantilesMembers.size(); i++) {
+ out << quantilesMembers[i].size() << '\t';
+ for (int j = 0; j < quantilesMembers[i].size(); j++) {
+ out << quantilesMembers[i][j] << '\t';
+ }
+ out << endl;
+ }
+
+ out.close();
+
+ exit(0);
+ }else {
+ m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
+ for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
exit(0);
- }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
+ }
}
+ //parent does its part
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, templateLines[0]->start, templateLines[0]->end);
+
//force parent to wait until all the processes are done
- for (int i=0;i<processors;i++) {
+ for (int i=0;i<(processors-1);i++) {
int temp = processIDS[i];
wait(&temp);
}
-
-#else
- findPairs(lines[0]->start, lines[0]->end);
-#endif
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesPairs");
- exit(1);
- }
-}
+ //get data created by processes
+ for (int i=0;i<(processors-1);i++) {
+ ifstream in;
+ string s = toString(processIDS[i]) + ".temp";
+ m->openInputFile(s, in);
+
+ vector< vector<float> > quan;
+ quan.resize(100);
+
+ //get quantiles
+ for (int h = 0; h < quan.size(); h++) {
+ int num;
+ in >> num;
+
+ m->gobble(in);
-/**************************************************************************************************/
+ vector<float> q; float w;
+ for (int j = 0; j < num; j++) {
+ in >> w;
+ q.push_back(w);
+ }
-void Pintail::createProcessesObserved() {
- try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- int process = 0;
- vector<int> processIDS;
-
- //loop through and create all the processes you want
- while (process != processors) {
- int pid = fork();
+ quan[h] = q;
+ m->gobble(in);
+ }
- if (pid > 0) {
- processIDS.push_back(pid);
- process++;
- }else if (pid == 0){
- calcObserved(lines[process]->start, lines[process]->end);
- exit(0);
- }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
- }
-
- //force parent to wait until all the processes are done
- for (int i=0;i<processors;i++) {
- int temp = processIDS[i];
- wait(&temp);
+
+ //save quan in quantiles
+ for (int j = 0; j < quan.size(); j++) {
+ //put all values of q[i] into quan[i]
+ for (int l = 0; l < quan[j].size(); l++) { quantilesMembers[j].push_back(quan[j][l]); }
+ //quantilesMembers[j].insert(quantilesMembers[j].begin(), quan[j].begin(), quan[j].end());
+ }
+
+ in.close();
+ m->mothurRemove(s);
}
-
-#else
- calcObserved(lines[0]->start, lines[0]->end);
+#else
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, 0, templateSeqs.size());
#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesObserved");
+ m->errorOut(e, "Pintail", "createProcessesQuan");
exit(1);
}
}
-
//***************************************************************************************************************
-
-void Pintail::createProcessesExpected() {
+vector< vector<float> > Pintail::readQuantiles() {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- int process = 0;
- vector<int> processIDS;
+ int num;
+ float ten, twentyfive, fifty, seventyfive, ninetyfive, ninetynine;
- //loop through and create all the processes you want
- while (process != processors) {
- int pid = fork();
+ vector< vector<float> > quan;
+ vector <float> temp; temp.resize(6, 0);
+
+ //to fill 0
+ quan.push_back(temp);
+
+ #ifdef USE_MPI
+
+ MPI_File inMPI;
+ MPI_Offset size;
+ MPI_Status status;
+
+ //char* inFileName = new char[quanfile.length()];
+ //memcpy(inFileName, quanfile.c_str(), quanfile.length());
+
+ char inFileName[1024];
+ strcpy(inFileName, quanfile.c_str());
+
+ MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
+ MPI_File_get_size(inMPI, &size);
+ //delete inFileName;
+
+
+ char* buffer = new char[size];
+ MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
+
+ string tempBuf = buffer;
+ if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size); }
+ istringstream iss (tempBuf,istringstream::in);
+ delete buffer;
+
+ //read version
+ string line = m->getline(iss); m->gobble(iss);
+
+ while(!iss.eof()) {
+ iss >> num >> ten >> twentyfive >> fifty >> seventyfive >> ninetyfive >> ninetynine;
- if (pid > 0) {
- processIDS.push_back(pid);
- process++;
- }else if (pid == 0){
- calcExpected(lines[process]->start, lines[process]->end);
- exit(0);
- }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
+ temp.clear();
+
+ temp.push_back(ten);
+ temp.push_back(twentyfive);
+ temp.push_back(fifty);
+ temp.push_back(seventyfive);
+ temp.push_back(ninetyfive);
+ temp.push_back(ninetynine);
+
+ quan.push_back(temp);
+
+ m->gobble(iss);
}
+
+ MPI_File_close(&inMPI);
- //force parent to wait until all the processes are done
- for (int i=0;i<processors;i++) {
- int temp = processIDS[i];
- wait(&temp);
+ #else
+
+ ifstream in;
+ m->openInputFile(quanfile, in);
+
+ //read version
+ string line = m->getline(in); m->gobble(in);
+
+ while(!in.eof()){
+
+ in >> num >> ten >> twentyfive >> fifty >> seventyfive >> ninetyfive >> ninetynine;
+
+ temp.clear();
+
+ temp.push_back(ten);
+ temp.push_back(twentyfive);
+ temp.push_back(fifty);
+ temp.push_back(seventyfive);
+ temp.push_back(ninetyfive);
+ temp.push_back(ninetynine);
+
+ quan.push_back(temp);
+
+ m->gobble(in);
}
+ in.close();
+ #endif
+
+ return quan;
-#else
- calcExpected(lines[0]->start, lines[0]->end);
-
-#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesExpected");
+ m->errorOut(e, "Pintail", "readQuantiles");
exit(1);
}
}
+//***************************************************************************************************************/
-/**************************************************************************************************/
-
-void Pintail::createProcessesDE() {
+void Pintail::printQuanFile(string file, string outputString) {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- int process = 0;
- vector<int> processIDS;
+
+ #ifdef USE_MPI
- //loop through and create all the processes you want
- while (process != processors) {
- int pid = fork();
+ MPI_File outQuan;
+ MPI_Status status;
- if (pid > 0) {
- processIDS.push_back(pid);
- process++;
- }else if (pid == 0){
- calcDE(lines[process]->start, lines[process]->end);
- exit(0);
- }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
- }
-
- //force parent to wait until all the processes are done
- for (int i=0;i<processors;i++) {
- int temp = processIDS[i];
- wait(&temp);
- }
-
-#else
- calcDE(lines[0]->start, lines[0]->end);
+ int pid;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
-#endif
+ int outMode=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());
+
+ if (pid == 0) {
+ MPI_File_open(MPI_COMM_SELF, FileName, outMode, MPI_INFO_NULL, &outQuan); //comm, filename, mode, info, filepointer
+
+ int length = outputString.length();
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
+
+ MPI_File_write(outQuan, buf, length, MPI_CHAR, &status);
+ delete buf;
+
+ MPI_File_close(&outQuan);
+ }
+
+ //delete FileName;
+ #else
+ ofstream outQuan;
+ m->openOutputFile(file, outQuan);
+
+ outQuan << outputString;
+
+ outQuan.close();
+ #endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesDE");
+ m->errorOut(e, "Pintail", "printQuanFile");
exit(1);
}
}
-//***************************************************************************************************************
+//***************************************************************************************************************/
+