#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 filename, string temp) { fastafile = filename; templateFile = temp; }
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 {
+ mothurOutEndLine();
+
for (int i = 0; i < querySeqs.size(); i++) {
int index = ceil(deviation[i]);
-
+
//is your DE value higher than the 95%
string chimera;
- if (DE[i] > quantiles[index][4]) { chimera = "Yes"; }
- else { chimera = "No"; }
+ 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";
for (int j = 0; j < obsDistance[i].size(); j++) { out << obsDistance[i][j] << '\t'; }
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
//break up file if needed
- int linesPerProcess = processors / numSeqs;
+ int linesPerProcess = numSeqs / processors ;
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//find breakup of sequences for all times we will Parallelize
templateLines.push_back(new linePair(0, templateSeqs.size()));
#endif
- distcalculator = new ignoreGaps();
-
-
+ distcalculator = new eachGapDist();
+ decalc = new DeCalculator();
+
+ //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()); }
+
+ decalc->setMask(seqMask);
+
+ //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(); }
+
+/*string o = "foronlinepintailpairs-eachgap";
+ofstream out7;
+openOutputFile(o, out7);
+
+for (int i = 0; i < bestfit.size(); i++) {
+ out7 << querySeqs[i]->getName() << endl;
+ out7 << querySeqs[i]->getUnaligned() << endl << endl;
+
+ out7 << bestfit[i]->getName() << endl;
+ out7 << bestfit[i]->getUnaligned() << endl << endl << 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(); }
+
+ //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();
+
+ //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]);
+ }
- //ex.align matches from wigeon
-/*for (int m = 0; m < templateSeqs.size(); m++) {
- if (templateSeqs[m]->getName() == "159481") { bestfit[17] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "100137") { bestfit[16] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "112956") { bestfit[15] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "102326") { bestfit[14] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "66229") { bestfit[13] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "206276") { bestfit[12] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "63607") { bestfit[11] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "7056") { bestfit[10] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "7088") { bestfit[9] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "17553") { bestfit[8] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "131723") { bestfit[7] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "69013") { bestfit[6] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "24543") { bestfit[5] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "27824") { bestfit[4] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "1456") { bestfit[3] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "1456") { bestfit[2] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "141312") { bestfit[1] = *(templateSeqs[m]); }
- if (templateSeqs[m]->getName() == "141312") { bestfit[0] = *(templateSeqs[m]); }
-
-
-}*/
+ 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]); }
+ createFilter(temp);
+
+ runFilter(querySeqs);
+ runFilter(templateSeqs);
+ runFilter(bestfit);
+ }
+
+
+ if (processors == 1) {
+
for (int j = 0; j < bestfit.size(); j++) {
- //chops off beginning and end of sequences so they both start and end with a base
- trimSeqs(querySeqs[j], bestfit[j], trimmed[j]);
+ decalc->trimSeqs(querySeqs[j], bestfit[j], trimmed[j]);
}
- mothurOut("Done."); mothurOutEndLine();
mothurOut("Finding window breaks... "); cout.flush();
for (int i = lines[0]->start; i < lines[0]->end; i++) {
it = trimmed[i].begin();
-cout << "trimmed = " << it->first << '\t' << it->second << endl;
- vector<int> win = findWindows(querySeqs[i], it->first, it->second, windowSizes[i]);
+ vector<int> win = decalc->findWindows(querySeqs[i], it->first, it->second, windowSizes[i], increment);
windowsForeachQuery[i] = win;
}
mothurOut("Done."); mothurOutEndLine();
}else { createProcessesSpots(); }
-
- //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 quantiles to a .prob file so that you can input them using the conservation parameter next time you run this command. Providing the .prob file will dramatically improve speed. "); cout.flush();
- probabilityProfile = calcFreq(templateSeqs);
- mothurOut("Done."); mothurOutEndLine();
- }else { probabilityProfile = readFreq(); }
-
- //make P into Q
- for (int i = 0; i < probabilityProfile.size(); i++) { probabilityProfile[i] = 1 - probabilityProfile[i]; }
- mothurOut("Done."); mothurOutEndLine();
if (processors == 1) {
mothurOut("Calculating observed distance... "); cout.flush();
for (int i = lines[0]->start; i < lines[0]->end; i++) {
- cout << querySeqs[i]->getName() << '\t' << bestfit[i].getName() << " windows = " << windowsForeachQuery[i].size() << " size = " << windowSizes[i] << endl;
- vector<float> obsi = calcObserved(querySeqs[i], bestfit[i], windowsForeachQuery[i], windowSizes[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 = findQav(windowsForeachQuery[i], windowSizes[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 = getCoef(obsDistance[i], Qav[i]);
- seqCoef.push_back(alpha);
+ 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 = calcExpected(Qav[i], seqCoef[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 = calcDE(obsDistance[i], expectedDistance[i]);
+ float de = decalc->calcDE(obsDistance[i], expectedDistance[i]);
DE[i] = de;
-
+
it = trimmed[i].begin();
- float dist = calcDist(querySeqs[i], bestfit[i], it->first, it->second);
+ float dist = decalc->calcDist(querySeqs[i], bestfit[i], it->first, it->second);
deviation[i] = dist;
}
mothurOut("Done."); mothurOutEndLine();
}
else { createProcesses(); }
-
//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(); }
+ 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) {
- quantiles = getQuantiles(0, templateSeqs.size());
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, 0, templateSeqs.size());
}else { createProcessesQuan(); }
+
- ofstream out4;
- string o = getRootName(templateFile) + "quan";
+ ofstream out4, out5;
+ string noOutliers, outliers;
- openOutputFile(o, out4);
+ 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";
+
+ /*openOutputFile(outliers, out4);
//adjust quantiles
- for (int i = 0; i < quantiles.size(); i++) {
- if (quantiles[i].size() == 0) {
+ 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++) {
- quantiles[i].push_back(0.0);
+ temp.push_back(0.0);
}
}else{
- sort(quantiles[i].begin(), quantiles[i].end());
+ sort(quantilesMembers[i].begin(), quantilesMembers[i].end(), compareQuanMembers);
- vector<float> temp;
//save 10%
- temp.push_back(quantiles[i][int(quantiles[i].size() * 0.10)]);
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.10)].score);
//save 25%
- temp.push_back(quantiles[i][int(quantiles[i].size() * 0.25)]);
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.25)].score);
//save 50%
- temp.push_back(quantiles[i][int(quantiles[i].size() * 0.5)]);
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.5)].score);
//save 75%
- temp.push_back(quantiles[i][int(quantiles[i].size() * 0.75)]);
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.75)].score);
//save 95%
- temp.push_back(quantiles[i][int(quantiles[i].size() * 0.95)]);
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.95)].score);
//save 99%
- temp.push_back(quantiles[i][int(quantiles[i].size() * 0.99)]);
+ temp.push_back(quantilesMembers[i][int(quantilesMembers[i].size() * 0.99)].score);
- quantiles[i] = temp;
}
//output quan value
out4 << i+1 << '\t';
- for (int u = 0; u < quantiles[i].size(); u++) { out4 << quantiles[i][u] << '\t'; }
+ for (int u = 0; u < temp.size(); u++) { out4 << temp[u] << '\t'; }
out4 << endl;
-
+
+ quantiles[i] = temp;
+
}
+ out4.close();*/
+
+ decalc->removeObviousOutliers(quantilesMembers, templateSeqs.size());
+
+ openOutputFile(noOutliers, out5);
+
+ //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();
+
}
-
+
//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", "getChimeras");
exit(1);
}
}
-//***************************************************************************************************************
-//num is query's spot in querySeqs
-void Pintail::trimSeqs(Sequence* query, Sequence subject, map<int, int>& trim) {
- try {
-
- string q = query->getAligned();
- string s = subject.getAligned();
-
- int front = 0;
- for (int i = 0; i < q.length(); i++) {
- if (isalpha(q[i]) && isalpha(s[i])) { front = i; break; }
- }
-
- int back = 0;
- for (int i = q.length(); i >= 0; i--) {
- if (isalpha(q[i]) && isalpha(s[i])) { back = i; break; }
- }
-
- trim[front] = back;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "trimSeqs");
- exit(1);
- }
-}
//***************************************************************************************************************
openInputFile(consfile, in);
vector<float> prob;
+ set<int> h = decalc->getPos(); //positions of bases in masking sequence
//read in probabilities and store in vector
int pos; float num;
in >> pos >> num;
- //do you want this spot
- prob.push_back(num);
+ 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);
+ }
gobble(in);
}
}
}
-//***************************************************************************************************************
-
-vector< vector<float> > Pintail::readQuantiles() {
- try {
-
- ifstream in;
- openInputFile(quanfile, in);
-
- vector< vector<float> > quan;
-
- int num; float ten, twentyfive, fifty, seventyfive, ninetyfive, ninetynine;
-
- while(!in.eof()){
-
- in >> num >> ten >> twentyfive >> fifty >> seventyfive >> ninetyfive >> ninetynine;
-
- vector <float> temp;
-
- 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);
-
- gobble(in);
- }
-
- in.close();
- return quan;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "readQuantiles");
- exit(1);
- }
-}
//***************************************************************************************************************
//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) {
+vector<Sequence*> Pintail::findPairs(int start, int end) {
try {
- vector<Sequence> seqsMatches; seqsMatches.resize(end-start);
+ vector<Sequence*> seqsMatches;
for(int i = start; i < end; i++){
float smallest = 10000.0;
Sequence query = *(querySeqs[i]);
-
+ Sequence* match;
+
for(int j = 0; j < templateSeqs.size(); j++){
Sequence temp = *(templateSeqs[j]);
float dist = distcalculator->getDist();
if (dist < smallest) {
- seqsMatches[i] = *(templateSeqs[j]);
+ match = templateSeqs[j];
smallest = dist;
}
}
+
+ //make copy so trimSeqs doesn't overtrim
+ Sequence* copy = new Sequence(match->getName(), match->getAligned());
+
+ seqsMatches.push_back(copy);
}
return seqsMatches;
}
}
-//***************************************************************************************************************
-//find the window breaks for each sequence - this is so you can move ahead by bases.
-vector<int> Pintail::findWindows(Sequence* query, int front, int back, int& size) {
+/**************************************************************************************************/
+
+void Pintail::createProcessesSpots() {
try {
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ int process = 0;
+ vector<int> processIDS;
- vector<int> win;
-
- int cutoff = back - front; //back - front
-
- //if window is set to default
- if (size == 0) { if (cutoff > 1200) { size = 300; }
- else{ size = (cutoff / 4); } }
- else if (size > (cutoff / 4)) {
- mothurOut("You have selected to large a window size for sequence " + query->getName() + ". I will choose an appropriate window size."); mothurOutEndLine();
- size = (cutoff / 4);
- }
-
- string seq = query->getAligned().substr(front, cutoff);
-
- //count bases
- int numBases = 0;
- for (int l = 0; l < seq.length(); l++) { if (isalpha(seq[l])) { numBases++; } }
-
- //save start of seq
- win.push_back(front);
-
- //move ahead increment bases at a time until all bases are in a window
- int countBases = 0;
- int totalBases = 0; //used to eliminate window of blanks at end of sequence
+ //loop through and create all the processes you want
+ while (process != processors) {
+ int pid = fork();
- seq = query->getAligned();
- for (int m = front; m < (back - size) ; m++) {
-
- //count number of bases you see
- if (isalpha(seq[m])) { countBases++; totalBases++; }
+ if (pid > 0) {
+ processIDS.push_back(pid);
+ process++;
+ }else if (pid == 0){
- //if you have seen enough bases to make a new window
- if (countBases >= increment) {
- win.push_back(m); //save spot in alignment
- countBases = 0; //reset bases you've seen in this window
- }
+ for (int j = lines[process]->start; j < lines[process]->end; j++) {
- //no need to continue if all your bases are in a window
- if (totalBases == numBases) { break; }
- }
-
- return win;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "findWindows");
- exit(1);
- }
-}
+ //chops off beginning and end of sequences so they both start and end with a base
+ map<int, int> trim;
-//***************************************************************************************************************
-vector<float> Pintail::calcObserved(Sequence* query, Sequence subject, vector<int> window, int size) {
- try {
-
- vector<float> temp;
-//cout << "query length = " << query->getAligned().length() << '\t' << " subject length = " << subject.getAligned().length() << endl;
- for (int m = 0; m < window.size(); m++) {
-
- string seqFrag = query->getAligned().substr(window[m], size);
- string seqFragsub = subject.getAligned().substr(window[m], size);
- //cout << "start point = " << window[m] << " end point = " << window[m]+size << endl;
- int diff = 0;
- for (int b = 0; b < seqFrag.length(); b++) {
- if (seqFrag[b] != seqFragsub[b]) { diff++; }
- }
-
- //percentage of mismatched bases
- float dist;
- dist = diff / (float) seqFrag.length() * 100;
-
- temp.push_back(dist);
- }
-
- return temp;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcObserved");
- exit(1);
- }
-}
-//***************************************************************************************************************
-float Pintail::calcDist(Sequence* query, Sequence subject, int front, int back) {
- try {
-
- //so you only look at the trimmed part of the sequence
- int cutoff = back - front;
-
- //from first startpoint with length back-front
- string seqFrag = query->getAligned().substr(front, cutoff);
- string seqFragsub = subject.getAligned().substr(front, cutoff);
-
- int diff = 0;
- for (int b = 0; b < seqFrag.length(); b++) {
- if (seqFrag[b] != seqFragsub[b]) { diff++; }
- }
-
- //percentage of mismatched bases
- float dist = diff / (float) seqFrag.length() * 100;
-
- return dist;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcDist");
- exit(1);
- }
-}
+ decalc->trimSeqs(querySeqs[j], bestfit[j], trim);
+ trimmed[j] = trim;
+
+ }
-//***************************************************************************************************************
-vector<float> Pintail::calcExpected(vector<float> qav, float coef) {
- try {
-
- //for each window
- vector<float> queryExpected;
-
- for (int m = 0; m < qav.size(); m++) {
+ 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();
- float expected = qav[m] * coef;
+ //write out data to file so parent can read it
+ ofstream out;
+ string s = toString(getpid()) + ".temp";
+ openOutputFile(s, out);
- queryExpected.push_back(expected);
- }
+ //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;
+ }
- return queryExpected;
+ //output windowSizes
+ for (int i = lines[process]->start; i < lines[process]->end; i++) {
+ out << windowSizes[i] << '\t';
+ }
+ out << endl;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcExpected");
- exit(1);
- }
-}
-//***************************************************************************************************************
-float Pintail::calcDE(vector<float> obs, vector<float> exp) {
- try {
-
- //for each window
- float sum = 0.0; //sum = sum from 1 to m of (oi-ei)^2
- for (int m = 0; m < obs.size(); m++) { sum += ((obs[m] - exp[m]) * (obs[m] - exp[m])); }
-
- float de = sqrt((sum / (obs.size() - 1)));
-
- return de;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcDE");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-
-vector<float> Pintail::calcFreq(vector<Sequence*> seqs) {
- try {
-
- vector<float> prob;
- string freqfile = getRootName(templateFile) + "prob";
- ofstream outFreq;
+ //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); }
+ }
- openOutputFile(freqfile, outFreq);
+ //force parent to wait until all the processes are done
+ for (int i=0;i<processors;i++) {
+ int temp = processIDS[i];
+ wait(&temp);
+ }
- //at each position in the sequence
- for (int i = 0; i < seqs[0]->getAligned().length(); i++) {
+ //get data created by processes
+ for (int i=0;i<processors;i++) {
+ ifstream in;
+ string s = toString(processIDS[i]) + ".temp";
+ openInputFile(s, in);
- vector<int> freq; freq.resize(4,0);
- int gaps = 0;
+ int size = lines[i]->end - lines[i]->start;
+
+ int count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int num;
+ in >> num;
- //find the frequency of each nucleotide
- for (int j = 0; j < seqs.size(); j++) {
-
- char value = seqs[j]->getAligned()[i];
+ 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;
- 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++; }
+ windowSizes[count] = num;
+ count++;
}
- //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;
- //subtract gaps to "ignore them"
- if ( (seqs.size() - gaps) == 0 ) { highFreq = 1.0; }
- else { highFreq = highest / (float) (seqs.size() - gaps); }
-
- float Pi;
- Pi = (highFreq - 0.25) / 0.75;
-
- //cannot have probability less than 0.
- if (Pi < 0) { Pi = 0.0; }
+ gobble(in);
- //saves this for later
- outFreq << i+1 << '\t' << Pi << endl;
+ count = lines[i]->start;
+ for (int m = 0; m < size; m++) {
+ int front, back;
+ in >> front >> back;
- prob.push_back(Pi);
- }
-
- outFreq.close();
-
- return prob;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "calcFreq");
- exit(1);
- }
-}
-//***************************************************************************************************************
-vector<float> Pintail::findQav(vector<int> window, int size) {
- try {
- vector<float> averages;
+ map<int, int> t;
- //for each window find average
- for (int m = 0; m < window.size(); m++) {
+ t[front] = back;
- float average = 0.0;
+ trimmed[count] = t;
+ count++;
- //while you are in the window for this sequence
- for (int j = window[m]; j < (window[m]+size); j++) { average += probabilityProfile[j]; }
-
- average = average / size;
-
- //save this windows average
- averages.push_back(average);
- }
-
- return averages;
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "findQav");
- exit(1);
- }
-}
+ gobble(in);
+ }
-//***************************************************************************************************************
-vector< vector<float> > Pintail::getQuantiles(int start, int end) {
- try {
- vector< vector<float> > quan;
-
- //percentage of mismatched pairs 1 to 100
- quan.resize(100);
-
-
- //for each sequence
- for(int i = start; i < end; i++){
-
- mothurOut("Processing template sequence " + toString(i)); mothurOutEndLine();
- Sequence* query = templateSeqs[i];
- //compare to every other sequence in template
- for(int j = 0; j < i; j++){
-
- Sequence subject = *(templateSeqs[j]);
-
- map<int, int> trim;
- trimSeqs(query, subject, trim);
-
- it = trim.begin();
- int front = it->first; int back = it->second;
-
- //reset window for each new comparison
- windowSizesTemplate[i] = window;
-
- vector<int> win = findWindows(query, front, back, windowSizesTemplate[i]);
-
- vector<float> obsi = calcObserved(query, subject, win, windowSizesTemplate[i]);
-
- vector<float> q = findQav(win, windowSizesTemplate[i]);
-
- float alpha = getCoef(obsi, q);
-
- vector<float> exp = calcExpected(q, alpha);
-
- float de = calcDE(obsi, exp);
-
- float dist = calcDist(query, subject, front, back);
-
- dist = ceil(dist);
-
- //dist-1 because vector indexes start at 0.
- quan[dist-1].push_back(de);
-
- }
+ in.close();
+ remove(s.c_str());
}
-
- return quan;
-
- }
- catch(exception& e) {
- errorOut(e, "Pintail", "findQav");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-float Pintail::getCoef(vector<float> obs, vector<float> qav) {
- try {
+
- //find average prob for this seqs windows
- float probAverage = 0.0;
- for (int j = 0; j < qav.size(); j++) { probAverage += qav[j]; }
- probAverage = probAverage / (float) qav.size();
-
- //find observed average
- float obsAverage = 0.0;
- for (int j = 0; j < obs.size(); j++) { obsAverage += obs[j]; }
- obsAverage = obsAverage / (float) obs.size();
-
+#else
+ 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]);
+ }
- float coef = obsAverage / probAverage;
-
- return coef;
+ 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", "getCoef");
+ errorOut(e, "Pintail", "createProcessesSpots");
exit(1);
}
}
/**************************************************************************************************/
-void Pintail::createProcessesSpots() {
+void Pintail::createProcessesPairs() {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
vector<int> processIDS;
- vector< vector<int> > win; win.resize(querySeqs.size());
//loop through and create all the processes you want
while (process != processors) {
}else if (pid == 0){
mothurOut("Finding pairs for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
- vector<Sequence> tempbest;
- tempbest = findPairs(lines[process]->start, lines[process]->end);
- int count = 0;
- for (int i = lines[process]->start; i < lines[process]->end; i++) {
- bestfit[i] = tempbest[count];
-
- //chops off beginning and end of sequences so they both start and end with a base
- trimSeqs(querySeqs[i], bestfit[i], trimmed[i]);
- count++;
- }
+ bestfit = findPairs(lines[process]->start, lines[process]->end);
mothurOut("Done finding pairs for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
- 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++) {
- vector<int> temp = findWindows(querySeqs[i], it->first, it->second, windowSizes[i]);
- win[i] = temp;
+ //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;
}
- mothurOut("Done finding window breaks for sequences " + toString(lines[process]->start) + " to " + toString(lines[process]->end)); mothurOutEndLine();
+ out.close();
exit(0);
}else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
wait(&temp);
}
- windowsForeachQuery = win;
+ //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
bestfit = 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
- trimSeqs(querySeqs[j], bestfit[j], j);
- }
-
- for (int i = lines[0]->start; i < lines[0]->end; i++) {
- it = trimmed[i].begin();
- map<int, int> win = findWindows(querySeqs[i], it->first, it->second, windowSizes[i]);
- windows[i] = win;
- }
-
#endif
}
catch(exception& e) {
- errorOut(e, "Pintail", "createProcessesSpots");
+ errorOut(e, "Pintail", "createProcessesPairs");
exit(1);
}
}
-
-
/**************************************************************************************************/
void Pintail::createProcesses() {
int process = 0;
vector<int> processIDS;
- vector< vector<float> > exp; exp.resize(querySeqs.size());
- vector<float> de; de.resize(querySeqs.size());
- vector< vector<float> > obs; obs.resize(querySeqs.size());
- vector<float> dev; dev.resize(querySeqs.size());
-
-
//loop through and create all the processes you want
while (process != processors) {
int pid = fork();
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 = calcObserved(querySeqs[i], bestfit[i], windowsForeachQuery[i], windowSizes[i]);
- obs[i] = obsi;
+ vector<float> obsi = decalc->calcObserved(querySeqs[i], bestfit[i], windowsForeachQuery[i], windowSizes[i]);
+ obsDistance[i] = obsi;
//calc Qav
- vector<float> q = findQav(windowsForeachQuery[i], windowSizes[i]);
+ vector<float> q = decalc->findQav(windowsForeachQuery[i], windowSizes[i], probabilityProfile);
//get alpha
- float alpha = getCoef(obsDistance[i], q);
+ float alpha = decalc->getCoef(obsDistance[i], q);
//find expected
- vector<float> exp = calcExpected(q, alpha);
+ vector<float> exp = decalc->calcExpected(q, alpha);
expectedDistance[i] = exp;
//get de and deviation
- float dei = calcDE(obsi, exp);
- de[i] = dei;
+ float dei = decalc->calcDE(obsi, exp);
+ DE[i] = dei;
it = trimmed[i].begin();
- float dist = calcDist(querySeqs[i], bestfit[i], it->first, it->second);
- dev[i] = dist;
+ 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);
}
- obsDistance = obs;
- expectedDistance = exp;
- DE = de;
- deviation = dev;
-
+ //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
mothurOut("Calculating observed distance... "); cout.flush();
for (int i = lines[0]->start; i < lines[0]->end; i++) {
- vector<float> obsi = calcObserved(querySeqs[i], bestfit[i], windows[i], windowSizes[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 = findQav(windows[i], windowSizes[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 = getCoef(obsDistance[i], Qav[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 = calcExpected(Qav[i], seqCoef[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 = calcDE(obsDistance[i], expectedDistance[i]);
+ float de = decalc->calcDE(obsDistance[i], expectedDistance[i]);
DE[i] = de;
it = trimmed[i].begin();
- float dist = calcDist(querySeqs[i], bestfit[i], it->first, it->second);
+ float dist = decalc->calcDist(querySeqs[i], bestfit[i], it->first, it->second);
deviation[i] = dist;
}
mothurOut("Done."); mothurOutEndLine();
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
vector<int> processIDS;
- vector< vector<float> > quan; quan.resize(100);
//loop through and create all the processes you want
while (process != processors) {
process++;
}else if (pid == 0){
- vector< vector<float> > q = getQuantiles(templateLines[process]->start, templateLines[process]->end);
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, templateLines[process]->start, templateLines[process]->end);
- for (int i = 0; i < q.size(); i++) {
- //put all values of q[i] into quan[i]
- quan[i].insert(quan[i].begin(), q[i].begin(), q[i].end());
- }
+ //write out data to file so parent can read it
+ ofstream out;
+ string s = toString(getpid()) + ".temp";
+ openOutputFile(s, out);
- for (int i = 0; i < quan.size(); i++) {
- cout << i+1 << '\t';
- for (int j = 0; j < quan[i].size(); j++) { cout << quan[i][j] << '\t'; }
- cout << endl;
+
+ //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());
+ }
- quantiles = quan;
#else
- quantiles = getQuantiles(0, templateSeqs.size());
+ quantilesMembers = decalc->getQuantiles(templateSeqs, windowSizesTemplate, window, probabilityProfile, increment, 0, templateSeqs.size());
#endif
}
catch(exception& e) {
projects / mothur.git / blobdiff