#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) {
- try {
- fastafile = name;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "Pintail");
- exit(1);
- }
-}
+Pintail::Pintail(string filename, string temp) { fastafile = filename; templateFile = temp; }
//***************************************************************************************************************
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]; }
+ for (int i = 0; i < bestfit.size(); i++) { delete bestfit[i]; }
}
catch(exception& e) {
errorOut(e, "Pintail", "~Pintail");
void Pintail::print(ostream& out) {
try {
- for (itCoef = DE.begin(); itCoef != DE.end(); itCoef++) {
+ mothurOutEndLine();
- out << itCoef->first->getName() << '\t' << itCoef->second << endl;
+ for (int i = 0; i < querySeqs.size(); i++) {
+
+ int index = ceil(deviation[i]);
+
+ //is your DE value higher than the 95%
+ string chimera;
+ if (quantiles[index][4] == 0.0) {
+ chimera = "Your template does not include sequences that provide quantile values at distance " + toString(index);
+ }else {
+ if (DE[i] > quantiles[index][4]) { chimera = "Yes"; }
+ else { chimera = "No"; }
+ }
+
+ out << querySeqs[i]->getName() << '\t' << "div: " << deviation[i] << "\tstDev: " << DE[i] << "\tchimera flag: " << chimera << endl;
+ if (chimera == "Yes") {
+ mothurOut(querySeqs[i]->getName() + "\tdiv: " + toString(deviation[i]) + "\tstDev: " + toString(DE[i]) + "\tchimera flag: " + chimera); mothurOutEndLine();
+ }
out << "Observed\t";
- itObsDist = obsDistance.find(itCoef->first);
- for (int i = 0; i < itObsDist->second.size(); i++) { out << itObsDist->second[i] << '\t'; }
+ for (int j = 0; j < obsDistance[i].size(); j++) { out << obsDistance[i][j] << '\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'; }
+ for (int m = 0; m < expectedDistance[i].size(); m++) { out << expectedDistance[i][m] << '\t'; }
out << endl;
}
-
}
catch(exception& e) {
errorOut(e, "Pintail", "print");
void Pintail::getChimeras() {
try {
- distCalculator = new ignoreGaps();
-
//read in query sequences and subject sequences
mothurOut("Reading sequences and template file... "); cout.flush();
querySeqs = readSeqs(fastafile);
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)); } }
-
- //calculate number of iters
- iters = (querySeqs[0]->getAligned().length() - window + 1) / increment;
+ obsDistance.resize(numSeqs);
+ expectedDistance.resize(numSeqs);
+ seqCoef.resize(numSeqs);
+ DE.resize(numSeqs);
+ Qav.resize(numSeqs);
+ bestfit.resize(numSeqs);
+ deviation.resize(numSeqs);
+ trimmed.resize(numSeqs);
+ windowSizes.resize(numSeqs, window);
+ windowSizesTemplate.resize(templateSeqs.size(), window);
+ windowsForeachQuery.resize(numSeqs);
+ h.resize(numSeqs);
+ quantiles.resize(100); //one for every percent mismatch
+ quantilesMembers.resize(100); //one for every percent mismatch
- int linesPerProcess = processors / numSeqs;
+ //break up file if needed
+ int linesPerProcess = numSeqs / processors ;
- //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)));
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ //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)));
+ }
+ //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));
}
- //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));
- }
+
+ //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());
+ }
+ }
+ #else
+ lines.push_back(new linePair(0, numSeqs));
+ templateLines.push_back(new linePair(0, templateSeqs.size()));
+ #endif
- //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();
- // itBest = bestfit.begin();
- // itBest++; itBest++;
-//cout << itBest->first->getName() << '\t' << itBest->second->getName() << endl;
- //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();
+ distcalculator = new eachGapDist();
+ decalc = new DeCalculator();
- //find P
- mothurOut("Calculating expected percentage differences for each sequence... "); cout.flush();
- vector<float> probabilityProfile = calcFreq(templateSeqs);
+ //if the user does enter a mask then you want to keep all the spots in the alignment
+ if (seqMask.length() == 0) { decalc->setAlignmentLength(querySeqs[0]->getAligned().length()); }
+ else { decalc->setAlignmentLength(seqMask.length()); }
- //make P into Q
- for (int i = 0; i < probabilityProfile.size(); i++) { probabilityProfile[i] = 1 - probabilityProfile[i]; }
-//cout << "after p into Q" << endl;
- //find Qav
- averageProbability = findQav(probabilityProfile);
-//cout << "after Qav" << endl;
- //find Coefficient - maps a sequence to its coefficient
- seqCoef = getCoef(averageProbability);
-//cout << "after coef" << endl;
- //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();
+ decalc->setMask(seqMask);
- //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();
+ //find pairs
+ if (processors == 1) {
+ mothurOut("Finding closest sequence in template to each sequence... "); cout.flush();
+ bestfit = findPairs(lines[0]->start, lines[0]->end);
+ mothurOut("Done."); mothurOutEndLine();
+ }else { createProcessesPairs(); }
-
- //free memory
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; }
- delete distCalculator;
+//string o = "closestmatch.eachgap.fasta";
+//ofstream out7;
+//openOutputFile(o, out7);
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "getChimeras");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
+//for (int i = 0; i < bestfit.size(); i++) {
+ //out7 << ">" << querySeqs[i]->getName() << "-"<< bestfit[i]->getName() << endl;
+ //out7 << bestfit[i]->getAligned() << endl;
+//}
+//out7.close();
+ //find P
+ mothurOut("Getting conservation... "); cout.flush();
+ if (consfile == "") {
+ 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, templateFile);
+ mothurOut("Done."); mothurOutEndLine();
+ }else { probabilityProfile = readFreq(); }
-vector<Sequence*> Pintail::readSeqs(string file) {
- try {
-
- ifstream in;
- openInputFile(file, in);
- vector<Sequence*> container;
+ //make P into Q
+ for (int i = 0; i < probabilityProfile.size(); i++) { probabilityProfile[i] = 1 - probabilityProfile[i]; } //cout << i << '\t' << probabilityProfile[i] << endl;
+ mothurOut("Done."); mothurOutEndLine();
- //read in seqs and store in vector
- while(!in.eof()){
- Sequence* current = new Sequence(in);
+ //mask sequences if the user wants to
+ if (seqMask != "") {
+ //mask querys
+ for (int i = 0; i < querySeqs.size(); i++) {
+ decalc->runMask(querySeqs[i]);
+ }
+
+ //mask templates
+ for (int i = 0; i < templateSeqs.size(); i++) {
+ decalc->runMask(templateSeqs[i]);
+ }
- if (current->getAligned() == "") { current->setAligned(current->getUnaligned()); }
+ for (int i = 0; i < bestfit.size(); i++) {
+ decalc->runMask(bestfit[i]);
+ }
+
+ }
+
+ if (filter) {
+ vector<Sequence*> temp = templateSeqs;
+ for (int i = 0; i < querySeqs.size(); i++) { temp.push_back(querySeqs[i]); }
- container.push_back(current);
+ createFilter(temp);
- gobble(in);
+ runFilter(querySeqs);
+ runFilter(templateSeqs);
+ runFilter(bestfit);
}
-
- 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++){
-
- float smallest = 10000.0;
- Sequence query = *(querySeqs[i]);
-
- for(int j = 0; j < templateSeqs.size(); j++){
-
- Sequence temp = *(templateSeqs[j]);
-
- distCalculator->calcDist(query, temp);
- float dist = distCalculator->getDist();
-
- if (dist < smallest) {
-
- bestfit[querySeqs[i]] = templateSeqs[j];
- smallest = dist;
- }
+
+
+ if (processors == 1) {
+
+ for (int j = 0; j < bestfit.size(); j++) {
+ decalc->trimSeqs(querySeqs[j], bestfit[j], trimmed[j]);
}
- }
+
+ mothurOut("Finding window breaks... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ it = trimmed[i].begin();
+ vector<int> win = decalc->findWindows(querySeqs[i], it->first, it->second, windowSizes[i], increment);
+ windowsForeachQuery[i] = win;
+ }
+ mothurOut("Done."); mothurOutEndLine();
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "findPairs");
- exit(1);
- }
-}
-//***************************************************************************************************************
-void Pintail::calcObserved(int start, int end) {
- try {
-
- for(int i = start; i < end; i++){
+ }else { createProcessesSpots(); }
- itBest = bestfit.find(querySeqs[i]);
- Sequence* query;
- Sequence* subject;
+ if (processors == 1) {
+
+ mothurOut("Calculating observed distance... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ vector<float> obsi = decalc->calcObserved(querySeqs[i], bestfit[i], windowsForeachQuery[i], windowSizes[i]);
+
+ obsDistance[i] = obsi;
+ }
+ mothurOut("Done."); mothurOutEndLine();
- if (itBest != bestfit.end()) {
- query = itBest->first;
- subject = itBest->second;
- }else{ mothurOut("Error in calcObserved"); mothurOutEndLine(); }
- vector<Sequence*> queryFragment;
- vector<Sequence*> subjectFragment;
+ mothurOut("Finding variability... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ vector<float> q = decalc->findQav(windowsForeachQuery[i], windowSizes[i], probabilityProfile);
+
+ Qav[i] = q;
+ }
+ mothurOut("Done."); mothurOutEndLine();
+
- int startpoint = 0;
- for (int m = 0; m < iters; m++) {
- string seqFrag = query->getAligned().substr(startpoint, window);
- Sequence* temp = new Sequence(query->getName(), seqFrag);
- queryFragment.push_back(temp);
-
- string seqFragsub = subject->getAligned().substr(startpoint, window);
- Sequence* tempsub = new Sequence(subject->getName(), seqFragsub);
- subjectFragment.push_back(tempsub);
-
- startpoint += increment;
+ mothurOut("Calculating alpha... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ float alpha = decalc->getCoef(obsDistance[i], Qav[i]);
+ seqCoef[i] = alpha;
+ }
+ mothurOut("Done."); mothurOutEndLine();
+
+
+ mothurOut("Calculating expected distance... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ vector<float> exp = decalc->calcExpected(Qav[i], seqCoef[i]);
+ expectedDistance[i] = exp;
}
+ mothurOut("Done."); mothurOutEndLine();
- for(int j = 0; j < iters; j++){
+ mothurOut("Finding deviation... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ float de = decalc->calcDE(obsDistance[i], expectedDistance[i]);
+ DE[i] = de;
- distCalculator->calcDist(*(queryFragment[j]), *(subjectFragment[j]));
- float dist = distCalculator->getDist();
-
- //convert to similiarity
- //dist = 1 - dist;
-
- //save oi
- obsDistance[query].push_back(dist);
+ it = trimmed[i].begin();
+ float dist = decalc->calcDist(querySeqs[i], bestfit[i], it->first, it->second);
+ deviation[i] = dist;
}
+ mothurOut("Done."); mothurOutEndLine();
+
+ }
+ else { createProcesses(); }
- //free memory
- for (int i = 0; i < queryFragment.size(); i++) { delete queryFragment[i]; }
- for (int i = 0; i < subjectFragment.size(); i++) { delete subjectFragment[i]; }
- }
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcObserved");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-void Pintail::calcExpected(int start, int end) {
- try {
+ //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 {
+
+ 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(); }
-
- //for each sequence
- for(int i = start; i < end; i++){
- itCoef = seqCoef.find(querySeqs[i]);
- float coef = itCoef->second;
+ ofstream out4, out5;
+ string noOutliers, outliers;
- //for each window
- vector<float> queryExpected;
- for (int m = 0; m < iters; m++) {
- float expected = averageProbability[m] * coef;
- queryExpected.push_back(expected);
+ if ((!filter) && (seqMask == "")) {
+ noOutliers = getRootName(templateFile) + "pintail.quan";
+ }else if ((filter) && (seqMask == "")) {
+ noOutliers = getRootName(templateFile) + "pintail.filtered.quan";
+ }else if ((!filter) && (seqMask != "")) {
+ noOutliers = getRootName(templateFile) + "pintail.masked.quan";
+ }else if ((filter) && (seqMask != "")) {
+ noOutliers = getRootName(templateFile) + "pintail.filtered.masked.quan";
}
+
+ //outliers = getRootName(templateFile) + "pintail.quanYESOUTLIERS";
- expectedDistance[querySeqs[i]] = queryExpected;
+ /*openOutputFile(outliers, out4);
- }
+ //adjust quantiles
+ for (int i = 0; i < quantilesMembers.size(); i++) {
+ vector<float> temp;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcExpected");
- exit(1);
- }
-}
-//***************************************************************************************************************
-void Pintail::calcDE(int start, int end) {
- try {
-
-
- //for each sequence
- for(int i = start; i < end; i++){
+ 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(), compareQuanMembers);
+
+ //save 10%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.10)].score);
+ //save 25%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.25)].score);
+ //save 50%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.5)].score);
+ //save 75%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.75)].score);
+ //save 95%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.95)].score);
+ //save 99%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.99)].score);
+
+ }
+
+ //output quan value
+ out4 << i+1 << '\t';
+ for (int u = 0; u < temp.size(); u++) { out4 << temp[u] << '\t'; }
+ out4 << endl;
+
+ quantiles[i] = temp;
+
+ }
- itObsDist = obsDistance.find(querySeqs[i]);
- vector<float> obs = itObsDist->second;
+ out4.close();*/
- itExpDist = expectedDistance.find(querySeqs[i]);
- vector<float> exp = itExpDist->second;
+ decalc->removeObviousOutliers(quantilesMembers, templateSeqs.size());
- //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])); }
+ openOutputFile(noOutliers, out5);
- float de = sqrt((sum / (iters - 1)));
+ //adjust quantiles
+ for (int i = 0; i < quantilesMembers.size(); i++) {
+ vector<float> temp;
+
+ 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(), compareQuanMembers);
+
+ //save 10%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.10)].score);
+ //save 25%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.25)].score);
+ //save 50%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.5)].score);
+ //save 75%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.75)].score);
+ //save 95%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.95)].score);
+ //save 99%
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.99)].score);
+
+ }
+
+ //output quan value
+ out5 << i+1 << '\t';
+ for (int u = 0; u < temp.size(); u++) { out5 << temp[u] << '\t'; }
+ out5 << endl;
+
+ quantiles[i] = temp;
+
+ }
+
+ mothurOut("Done."); mothurOutEndLine();
- DE[querySeqs[i]] = de;
}
-
+
+ //free memory
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; }
+ for (int i = 0; i < templateLines.size(); i++) { delete templateLines[i]; }
+
+ delete distcalculator;
+ delete decalc;
}
catch(exception& e) {
- errorOut(e, "Pintail", "calcDE");
+ errorOut(e, "Pintail", "getChimeras");
exit(1);
}
}
//***************************************************************************************************************
-vector<float> Pintail::calcFreq(vector<Sequence*> seqs) {
+vector<float> Pintail::readFreq() {
try {
-
+
+ ifstream in;
+ openInputFile(consfile, in);
+
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++) {
+ //read in probabilities and store in vector
+ int pos; float num;
- vector<int> freq; freq.resize(4,0);
+ while(!in.eof()){
- //find the frequency of each nucleotide
- for (int j = 0; j < seqs.size(); j++) {
-
- char value = seqs[j]->getAligned()[i];
+ in >> pos >> num;
- 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]++; }
- }
+ if (h.count(pos) > 0) {
+ float Pi;
+ Pi = (num - 0.25) / 0.75;
- //find base with highest frequency
- int highest = 0;
- for (int m = 0; m < freq.size(); m++) { if (freq[m] > highest) { highest = freq[m]; } }
-
- float highFreq = highest / (float) seqs.size();
- float Pi;
- //if this is not a gap column
- if (highFreq != 0) { Pi = (highFreq - 0.25) / 0.75; }
- else { Pi = 0; }
-
- prob.push_back(Pi);
+ //cannot have probability less than 0.
+ if (Pi < 0) { Pi = 0.0; }
+
+ //do you want this spot
+ prob.push_back(Pi);
+ }
+ gobble(in);
}
+ in.close();
return prob;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcFreq");
- exit(1);
- }
-}
-//***************************************************************************************************************
-vector<float> Pintail::findQav(vector<float> prob) {
- try {
- vector<float> averages;
-
- //for each window find average
- int startpoint = 0;
- for (int m = 0; m < iters; m++) {
-
- float average = 0.0;
- for (int i = startpoint; i < (startpoint+window); i++) { average += prob[i]; }
-
- average = average / window;
-
- //save this windows average
- averages.push_back(average);
-
- startpoint += increment;
- }
- return averages;
}
catch(exception& e) {
- errorOut(e, "Pintail", "findQav");
+ 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
+vector<Sequence*> Pintail::findPairs(int start, int end) {
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();
+ vector<Sequence*> seqsMatches;
- //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;
+ for(int i = start; i < end; i++){
+
+ float smallest = 10000.0;
+ Sequence query = *(querySeqs[i]);
+ Sequence* match;
- //find observed average
- float obsAverage = 0.0;
- for (int i = 0; i < temp.size(); i++) { obsAverage += temp[i]; }
- obsAverage = obsAverage / (float) temp.size();
+ for(int j = 0; j < templateSeqs.size(); j++){
+
+ Sequence temp = *(templateSeqs[j]);
+
+ distcalculator->calcDist(query, temp);
+ float dist = distcalculator->getDist();
+
+ if (dist < smallest) {
+ match = templateSeqs[j];
+ smallest = dist;
+ }
+ }
- float coef = obsAverage / probAverage;
+ //make copy so trimSeqs doesn't overtrim
+ Sequence* copy = new Sequence(match->getName(), match->getAligned());
- //save this sequences coefficient
- coefs[tempSeq] = coef;
+ seqsMatches.push_back(copy);
}
-
- return coefs;
+ return seqsMatches;
+
}
catch(exception& e) {
- errorOut(e, "Pintail", "getCoef");
+ errorOut(e, "Pintail", "findPairs");
exit(1);
}
}
/**************************************************************************************************/
-void Pintail::createProcessesPairs() {
+void Pintail::createProcessesSpots() {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
processIDS.push_back(pid);
process++;
}else if (pid == 0){
- findPairs(lines[process]->start, lines[process]->end);
+
+ for (int j = lines[process]->start; j < lines[process]->end; j++) {
+
+ //chops off beginning and end of sequences so they both start and end with a base
+ map<int, int> trim;
+
+ decalc->trimSeqs(querySeqs[j], bestfit[j], trim);
+ trimmed[j] = trim;
+
+ }
+
+ mothurOut("Finding window breaks for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ it = trimmed[i].begin();
+ windowsForeachQuery[i] = decalc->findWindows(querySeqs[i], it->first, it->second, windowSizes[i], increment);
+ }
+ mothurOut("Done finding window breaks for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+
+ //write out data to file so parent can read it
+ ofstream out;
+ string s = toString(getpid()) + ".temp";
+ openOutputFile(s, out);
+
+ //output windowsForeachQuery
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << windowsForeachQuery[i].size() << '\t';
+ for (int j = 0; j < windowsForeachQuery[i].size(); j++) {
+ out << windowsForeachQuery[i][j] << '\t';
+ }
+ out << endl;
+ }
+
+ //output windowSizes
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << windowSizes[i] << '\t';
+ }
+ out << endl;
+
+ //output trimmed values
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ it = trimmed[i].begin();
+ out << it->first << '\t' << it->second << endl;
+ }
+ out.close();
+
exit(0);
}else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
}
wait(&temp);
}
+ //get data created by processes
+ for (int i=0;i<processors;i++) {
+ ifstream in;
+ string s = toString(processIDS[i]) + ".temp";
+ openInputFile(s, in);
+
+ int size = lines[i]->end - lines[i]->start;
+
+ int count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int num;
+ in >> num;
+
+ vector<int> win; int w;
+ for (int j = 0; j < num; j++) {
+ in >> w;
+ win.push_back(w);
+ }
+
+ windowsForeachQuery[count] = win;
+ count++;
+ gobble(in);
+ }
+
+ gobble(in);
+ count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int num;
+ in >> num;
+
+ windowSizes[count] = num;
+ count++;
+ }
+
+ gobble(in);
+
+ count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int front, back;
+ in >> front >> back;
+
+ map<int, int> t;
+
+ t[front] = back;
+
+ trimmed[count] = t;
+ count++;
+
+ gobble(in);
+ }
+
+
+ in.close();
+ remove(s.c_str());
+ }
+
+
#else
- findPairs(lines[0]->start, lines[0]->end);
+ for (int j = 0; j < bestfit.size(); j++) {
+ //chops off beginning and end of sequences so they both start and end with a base
+ decalc->trimSeqs(querySeqs[j], bestfit[j], trimmed[j]);
+ }
+
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ it = trimmed[i].begin();
+ vector<int> win = decalc->findWindows(querySeqs[i], it->first, it->second, windowSizes[i], increment);
+ windowsForeachQuery[i] = win;
+ }
#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesPairs");
+ errorOut(e, "Pintail", "createProcessesSpots");
exit(1);
}
}
-
/**************************************************************************************************/
-void Pintail::createProcessesObserved() {
+void Pintail::createProcessesPairs() {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
processIDS.push_back(pid);
process++;
}else if (pid == 0){
- calcObserved(lines[process]->start, lines[process]->end);
+
+ mothurOut("Finding pairs for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+ bestfit = findPairs(lines[process]->start, lines[process]->end);
+ mothurOut("Done finding pairs for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+
+ //write out data to file so parent can read it
+ ofstream out;
+ string s = toString(getpid()) + ".temp";
+ openOutputFile(s, out);
+
+ //output range and size
+ out << bestfit.size() << endl;
+
+ //output pairs
+ for (int i = 0; i < bestfit.size(); i++) {
+ out << ">" << bestfit[i]->getName() << endl << bestfit[i]->getAligned() << endl;
+ }
+ out.close();
+
exit(0);
}else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
}
wait(&temp);
}
+ //get data created by processes
+ for (int i=0;i<processors;i++) {
+ ifstream in;
+ string s = toString(processIDS[i]) + ".temp";
+ openInputFile(s, in);
+
+ int size;
+ in >> size; gobble(in);
+
+ //get pairs
+ int count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ Sequence* temp = new Sequence(in);
+ bestfit[count] = temp;
+
+ count++;
+ gobble(in);
+ }
+
+ in.close();
+ remove(s.c_str());
+ }
+
+
#else
- calcObserved(lines[0]->start, lines[0]->end);
-
+ bestfit = findPairs(lines[0]->start, lines[0]->end);
#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesObserved");
+ errorOut(e, "Pintail", "createProcessesPairs");
exit(1);
}
}
+/**************************************************************************************************/
-//***************************************************************************************************************
-
-void Pintail::createProcessesExpected() {
+void Pintail::createProcesses() {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
processIDS.push_back(pid);
process++;
}else if (pid == 0){
- calcExpected(lines[process]->start, lines[process]->end);
+
+ mothurOut("Calculating observed, expected and de values for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+
+ vector<float> obsi = decalc->calcObserved(querySeqs[i], bestfit[i], windowsForeachQuery[i], windowSizes[i]);
+ obsDistance[i] = obsi;
+
+ //calc Qav
+ vector<float> q = decalc->findQav(windowsForeachQuery[i], windowSizes[i], probabilityProfile);
+
+ //get alpha
+ float alpha = decalc->getCoef(obsDistance[i], q);
+
+ //find expected
+ vector<float> exp = decalc->calcExpected(q, alpha);
+ expectedDistance[i] = exp;
+
+ //get de and deviation
+ float dei = decalc->calcDE(obsi, exp);
+ DE[i] = dei;
+
+ it = trimmed[i].begin();
+ float dist = decalc->calcDist(querySeqs[i], bestfit[i], it->first, it->second);
+ deviation[i] = dist;
+ }
+ mothurOut("Done calculating observed, expected and de values for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+
+ //write out data to file so parent can read it
+ ofstream out;
+ string s = toString(getpid()) + ".temp";
+ openOutputFile(s, out);
+
+ int size = lines[process]->end - lines[process]->start;
+ out << size << endl;
+
+ //output observed distances
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << obsDistance[i].size() << '\t';
+ for (int j = 0; j < obsDistance[i].size(); j++) {
+ out << obsDistance[i][j] << '\t';
+ }
+ out << endl;
+ }
+
+
+ //output expected distances
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << expectedDistance[i].size() << '\t';
+ for (int j = 0; j < expectedDistance[i].size(); j++) {
+ out << expectedDistance[i][j] << '\t';
+ }
+ out << endl;
+ }
+
+
+ //output de values
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << DE[i] << '\t';
+ }
+ out << endl;
+
+ //output de values
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << deviation[i] << '\t';
+ }
+ out << endl;
+
+ out.close();
+
exit(0);
}else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
}
wait(&temp);
}
+ //get data created by processes
+ for (int i=0;i<processors;i++) {
+ ifstream in;
+ string s = toString(processIDS[i]) + ".temp";
+ openInputFile(s, in);
+
+ int size;
+ in >> size; gobble(in);
+
+ //get observed distances
+ int count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int num;
+ in >> num;
+
+ vector<float> obs; float w;
+ for (int j = 0; j < num; j++) {
+ in >> w;
+ obs.push_back(w);
+ }
+
+ obsDistance[count] = obs;
+ count++;
+ gobble(in);
+ }
+
+ gobble(in);
+
+ //get expected distances
+ count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int num;
+ in >> num;
+
+ vector<float> exp; float w;
+ for (int j = 0; j < num; j++) {
+ in >> w;
+ exp.push_back(w);
+ }
+
+ expectedDistance[count] = exp;
+ count++;
+ gobble(in);
+ }
+
+ gobble(in);
+
+ count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ float num;
+ in >> num;
+
+ DE[count] = num;
+ count++;
+ }
+
+ gobble(in);
+
+ count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ float num;
+ in >> num;
+
+ deviation[count] = num;
+ count++;
+ }
+
+ in.close();
+ remove(s.c_str());
+ }
+
+
#else
- calcExpected(lines[0]->start, lines[0]->end);
+ mothurOut("Calculating observed distance... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ vector<float> obsi = decalc->calcObserved(querySeqs[i], bestfit[i], windowsForeachQuery[i], windowSizes[i]);
+ obsDistance[i] = obsi;
+ }
+ mothurOut("Done."); mothurOutEndLine();
+
+
+
+ mothurOut("Finding variability... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ vector<float> q = decalc->findQav(windowsForeachQuery[i], windowSizes[i], probabilityProfile);
+ Qav[i] = q;
+ }
+ mothurOut("Done."); mothurOutEndLine();
+
+
+
+ mothurOut("Calculating alpha... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ float alpha = decalc->getCoef(obsDistance[i], Qav[i]);
+ seqCoef.push_back(alpha);
+ }
+ mothurOut("Done."); mothurOutEndLine();
+
+
+
+ mothurOut("Calculating expected distance... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ vector<float> exp = decalc->calcExpected(Qav[i], seqCoef[i]);
+ expectedDistance[i] = exp;
+ }
+ mothurOut("Done."); mothurOutEndLine();
+
+
+
+ mothurOut("Finding deviation... "); cout.flush();
+ for (int i = lines[0]->start; i < lines[0]->end; i++) {
+ float de = decalc->calcDE(obsDistance[i], expectedDistance[i]);
+ DE[i] = de;
+
+ it = trimmed[i].begin();
+ float dist = decalc->calcDist(querySeqs[i], bestfit[i], it->first, it->second);
+ deviation[i] = dist;
+ }
+ mothurOut("Done."); mothurOutEndLine();
#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesExpected");
+ errorOut(e, "Pintail", "createProcesses");
exit(1);
}
}
+
/**************************************************************************************************/
-void Pintail::createProcessesDE() {
+void Pintail::createProcessesQuan() {
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();
processIDS.push_back(pid);
process++;
}else if (pid == 0){
- calcDE(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";
+ 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].score << '\t' << quantilesMembers[i][j].member1 << '\t' << quantilesMembers[i][j].member2 << '\t';
+ }
+ out << endl;
+ }
+
+ out.close();
+
exit(0);
}else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
}
int temp = processIDS[i];
wait(&temp);
}
+
+ //get data created by processes
+ for (int i=0;i<processors;i++) {
+ ifstream in;
+ string s = toString(processIDS[i]) + ".temp";
+ openInputFile(s, in);
+
+ vector< vector<quanMember> > quan;
+ quan.resize(100);
+
+ //get quantiles
+ for (int m = 0; m < quan.size(); m++) {
+ int num;
+ in >> num;
+
+ gobble(in);
+
+ vector<quanMember> q; float w; int b, n;
+ for (int j = 0; j < num; j++) {
+ in >> w >> b >> n;
+ //cout << w << '\t' << b << '\t' n << endl;
+ quanMember newMember(w, b, n);
+ q.push_back(newMember);
+ }
+//cout << "here" << endl;
+ quan[m] = q;
+//cout << "now here" << endl;
+ gobble(in);
+ }
+
+
+ //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();
+ remove(s.c_str());
+ }
#else
- calcDE(lines[0]->start, lines[0]->end);
-
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, 0, templateSeqs.size());
#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesDE");
+ errorOut(e, "Pintail", "createProcessesQuan");
exit(1);
}
}
+
//***************************************************************************************************************
projects / mothur.git / blobdiff