*/
#include "maligner.h"
-#include "kmerdb.hpp"
-#include "blastdb.hpp"
-/***********************************************************************/
-Maligner::Maligner(vector<Sequence*> temp, int num, int match, int misMatch, float div, int ms, int minCov, string mode, Database* dataLeft, Database* dataRight) :
- db(temp), numWanted(num), matchScore(match), misMatchPenalty(misMatch), minDivR(div), minSimilarity(ms), minCoverage(minCov), searchMethod(mode), databaseLeft(dataLeft), databaseRight(dataRight) { m = MothurOut::getInstance(); }
+/***********************************************************************/ //int num, int match, int misMatch, , string mode, Database* dataLeft, Database* dataRight
+Maligner::Maligner(vector<Sequence*> temp, int match, int misMatch, float div, int ms, int minCov) : db(temp), matchScore(match), misMatchPenalty(misMatch), minDivR(div), minSimilarity(ms), minCoverage(minCov) {
+ //numWanted(num), , searchMethod(mode), databaseLeft(dataLeft), databaseRight(dataRight)
+
+ m = MothurOut::getInstance();
+
+}
/***********************************************************************/
string Maligner::getResults(Sequence* q, DeCalculator* decalc) {
try {
string chimera;
- if (searchMethod == "distance") {
- //find closest seqs to query in template - returns copies of seqs so trim does not destroy - remember to deallocate
- refSeqs = decalc->findClosest(query, db, numWanted, indexes);
- }else if (searchMethod == "blast") {
- refSeqs = getBlastSeqs(query, numWanted); //fills indexes
- }else if (searchMethod == "kmer") {
- refSeqs = getKmerSeqs(query, numWanted); //fills indexes
- }else { m->mothurOut("not valid search."); exit(1); } //should never get here
-
- if (m->control_pressed) { return chimera; }
+ //copy refSeqs so that filter does not effect original
+ for(int i = 0; i < db.size(); i++) {
+ Sequence* newSeq = new Sequence(db[i]->getName(), db[i]->getAligned());
+ refSeqs.push_back(newSeq);
+ }
refSeqs = minCoverageFilter(refSeqs);
-
+
if (refSeqs.size() < 2) {
for (int i = 0; i < refSeqs.size(); i++) { delete refSeqs[i]; }
percentIdenticalQueryChimera = 0.0;
}
int chimeraPenalty = computeChimeraPenalty();
-
+
//fills outputResults
chimera = chimeraMaligner(chimeraPenalty, decalc);
//trims seqs to first non gap char in all seqs and last non gap char in all seqs
spotMap = decalc->trimSeqs(query, refSeqs);
+
+ //you trimmed the whole sequence, skip
+ if (query->getAligned() == "") { return "no"; }
vector<Sequence*> temp = refSeqs;
temp.push_back(query);
-
+
verticalFilter(temp);
-//for (int i = 0; i < temp.size(); i++) { cout << temp[i]->getName() << '\n' << temp[i]->getAligned() << endl; } return "no";
+
+ //for (int i = 0; i < refSeqs.size(); i++) { cout << refSeqs[i]->getName() << endl << refSeqs[i]->getAligned() << endl; }
vector< vector<score_struct> > matrix = buildScoreMatrix(query->getAligned().length(), refSeqs.size()); //builds and initializes
fillScoreMatrix(matrix, refSeqs, chimeraPenalty);
- vector<score_struct> path = extractHighestPath(matrix);
-
- if (m->control_pressed) { return chimera; }
+ vector<trace_struct> trace = extractHighestPath(matrix);
+
+ //cout << "traces\n";
+ //for(int i=0;i<trace.size();i++){
+ // cout << trace[i].col << '\t' << trace[i].oldCol << '\t' << refSeqs[trace[i].row]->getName() << endl;
+ //}
- vector<trace_struct> trace = mapTraceRegionsToAlignment(path, refSeqs);
-
if (trace.size() > 1) { chimera = "yes"; }
- else { chimera = "no"; }
+ else { chimera = "no"; return chimera; }
- int traceStart = path[0].col;
- int traceEnd = path[path.size()-1].col;
+ int traceStart = trace[0].col;
+ int traceEnd = trace[trace.size()-1].oldCol;
string queryInRange = query->getAligned();
queryInRange = queryInRange.substr(traceStart, (traceEnd-traceStart+1));
-
- string chimeraSeq = constructChimericSeq(trace, refSeqs);
-
- percentIdenticalQueryChimera = computePercentID(queryInRange, chimeraSeq);
if (m->control_pressed) { return chimera; }
results temp;
temp.parent = refSeqs[seqIndex]->getName();
- temp.parentAligned = db[indexes[seqIndex]]->getAligned();
+ temp.parentAligned = db[seqIndex]->getAligned();
temp.nastRegionStart = spotMap[regionStart];
temp.nastRegionEnd = spotMap[regionEnd];
temp.regionStart = regionStart;
temp.queryToParent = computePercentID(queryInRange, parentInRange);
temp.divR = (percentIdenticalQueryChimera / temp.queryToParent);
-
+
string queryInRegion = query->getAligned();
queryInRegion = queryInRegion.substr(regionStart, (regionEnd-regionStart+1));
parentInRegion = parentInRegion.substr(regionStart, (regionEnd-regionStart+1));
temp.queryToParentLocal = computePercentID(queryInRegion, parentInRegion);
-
+
+ //cout << temp.parent << '\t' << "NAST:" << temp.nastRegionStart << '-' << temp.nastRegionEnd << " G:" << temp.queryToParent << " L:" << temp.queryToParentLocal << endl;
outputResults.push_back(temp);
}
-
+
return chimera;
}
catch(exception& e) {
//if coverage above minimum
if (coverage > minCoverage) {
newRefs.push_back(ref[i]);
+ }else {
+ delete ref[i];
}
}
try {
int numAllowable = ((1.0 - (1.0/minDivR)) * query->getNumBases());
-
+
+// if(numAllowable < 1){ numAllowable = 1; }
+
int penalty = int(numAllowable + 1) * misMatchPenalty;
-
+
return penalty;
}
vector< vector<score_struct> > Maligner::buildScoreMatrix(int cols, int rows) {
try{
- vector< vector<score_struct> > m; m.resize(rows);
+ vector< vector<score_struct> > m(rows);
- for (int i = 0; i < m.size(); i++) {
+ for (int i = 0; i < rows; i++) {
for (int j = 0; j < cols; j++) {
//initialize each cell
exit(1);
}
}
+
//***************************************************************************************************************
+
void Maligner::fillScoreMatrix(vector<vector<score_struct> >& ms, vector<Sequence*> seqs, int penalty) {
try{
//are you both gaps?
if ((!isalpha(queryAligned[0])) && (!isalpha(subjectAligned[0]))) {
ms[i][0].score = 0;
- }else if (queryAligned[0] == subjectAligned[0]) {
+// ms[i][0].mismatches = 0;
+ }else if (queryAligned[0] == subjectAligned[0]) { //|| subjectAligned[0] == 'N')
ms[i][0].score = matchScore;
+// ms[i][0].mismatches = 0;
}else{
ms[i][0].score = 0;
+// ms[i][0].mismatches = 1;
}
}
if ((!isalpha(queryAligned[j])) && (!isalpha(subjectAligned[j]))) {
//leave the same
}else if ((toupper(queryAligned[j]) == 'N') || (toupper(subjectAligned[j]) == 'N')) {
+ //matchMisMatchScore = matchScore;
//leave the same
}else if (queryAligned[j] == subjectAligned[j]) {
matchMisMatchScore = matchScore;
+// ms[i][j].mismatches = ms[i][j-1].mismatches;
}else if (queryAligned[j] != subjectAligned[j]) {
matchMisMatchScore = misMatchPenalty;
+// ms[i][j].mismatches = ms[i][j-1].mismatches + 1;
}
//compute score based on previous columns scores
}
}
+ /* for(int i=0;i<numRows;i++){
+ cout << seqs[i]->getName();
+ for(int j=0;j<numCols;j++){
+ cout << '\t' << ms[i][j].mismatches;
+ }
+ cout << endl;
+ }
+ cout << endl;*/
+ /*cout << numRows << '\t' << numCols << endl;
+ for(int i=0;i<numRows;i++){
+ cout << seqs[i]->getName() << endl << seqs[i]->getAligned() << endl << endl;
+ if ((seqs[i]->getName() == "S000003470") || (seqs[i]->getName() == "S000383265") || (seqs[i]->getName() == "7000004128191054")) {
+ for(int j=0;j<numCols;j++){
+ cout << '\t' << ms[i][j].score;
+ }
+ cout << endl;
+ }
+ }
+ cout << endl;*/
+ /*for(int i=0;i<numRows;i++){
+ cout << seqs[i]->getName();
+ for(int j=0;j<numCols;j++){
+ cout << '\t' << ms[i][j].prev;
+ }
+ cout << endl;
+ }*/
+
+
}
catch(exception& e) {
m->errorOut(e, "Maligner", "fillScoreMatrix");
exit(1);
}
}
+
//***************************************************************************************************************
-vector<score_struct> Maligner::extractHighestPath(vector<vector<score_struct> > ms) {
+
+vector<trace_struct> Maligner::extractHighestPath(vector<vector<score_struct> > ms) {
try {
+
//get matrix dimensions
int numCols = query->getAligned().length();
int numRows = ms.size();
//find highest score scoring matrix
- score_struct highestStruct;
+ vector<score_struct> highestStruct;
int highestScore = 0;
for (int i = 0; i < numRows; i++) {
for (int j = 0; j < numCols; j++) {
if (ms[i][j].score > highestScore) {
highestScore = ms[i][j].score;
- highestStruct = ms[i][j];
+ highestStruct.resize(0);
+ highestStruct.push_back(ms[i][j]);
+ }
+ else if(ms[i][j].score == highestScore){
+ highestStruct.push_back(ms[i][j]);
}
}
}
-
- vector<score_struct> path;
+
+ //cout << endl << highestScore << '\t' << highestStruct.size() << '\t' << highestStruct[0].row << endl;
- int rowIndex = highestStruct.row;
- int pos = highestStruct.col;
- int score = highestStruct.score;
+ vector<trace_struct> maxTrace;
+ double maxPercentIdenticalQueryAntiChimera = 0;
- while (pos >= 0 && score > 0) {
- score_struct temp = ms[rowIndex][pos];
- score = temp.score;
+ for(int i=0;i<highestStruct.size();i++){
- if (score > 0) { path.push_back(temp); }
+ vector<score_struct> path;
+
+ int rowIndex = highestStruct[i].row;
+ int pos = highestStruct[i].col;
+ int score = highestStruct[i].score;
+
+ while (pos >= 0 && score > 0) {
+ score_struct temp = ms[rowIndex][pos];
+ score = temp.score;
+
+ if (score > 0) { path.push_back(temp); }
+
+ rowIndex = temp.prev;
+ pos--;
+ }
+
+ reverse(path.begin(), path.end());
+
+ vector<trace_struct> trace = mapTraceRegionsToAlignment(path, refSeqs);
+
+ //cout << "traces\n";
+ //for(int j=0;j<trace.size();j++){
+ // cout << trace[j].col << '\t' << trace[j].oldCol << '\t' << refSeqs[trace[j].row]->getName() << endl;
+ //}
+
+ int traceStart = path[0].col;
+ int traceEnd = path[path.size()-1].col;
+// cout << "traceStart/End\t" << traceStart << '\t' << traceEnd << endl;
+
+ string queryInRange = query->getAligned();
+ queryInRange = queryInRange.substr(traceStart, (traceEnd-traceStart+1));
+// cout << "here" << endl;
+ string chimeraSeq = constructChimericSeq(trace, refSeqs);
+ string antiChimeraSeq = constructAntiChimericSeq(trace, refSeqs);
+
+ percentIdenticalQueryChimera = computePercentID(queryInRange, chimeraSeq);
+ double percentIdenticalQueryAntiChimera = computePercentID(queryInRange, antiChimeraSeq);
+// cout << i << '\t' << percentIdenticalQueryChimera << '\t' << percentIdenticalQueryAntiChimera << endl;
- rowIndex = temp.prev;
- pos--;
+ if(percentIdenticalQueryAntiChimera > maxPercentIdenticalQueryAntiChimera){
+ maxPercentIdenticalQueryAntiChimera = percentIdenticalQueryAntiChimera;
+ maxTrace = trace;
+ }
}
-
- reverse(path.begin(), path.end());
+// cout << maxPercentIdenticalQueryAntiChimera << endl;
+ return maxTrace;
- return path;
}
catch(exception& e) {
exit(1);
}
}
+
//***************************************************************************************************************
+
vector<trace_struct> Maligner::mapTraceRegionsToAlignment(vector<score_struct> path, vector<Sequence*> seqs) {
try {
vector<trace_struct> trace;
temp.row = region_index;
trace.push_back(temp);
+// cout << endl;
+// cout << trace.size() << endl;
+// for(int i=0;i<trace.size();i++){
+// cout << seqs[trace[i].row]->getName() << endl;
+// }
+// cout << endl;
+
return trace;
}
exit(1);
}
}
+
//***************************************************************************************************************
+
string Maligner::constructChimericSeq(vector<trace_struct> trace, vector<Sequence*> seqs) {
try {
string chimera = "";
for (int i = 0; i < trace.size(); i++) {
+// cout << i << '\t' << trace[i].row << '\t' << trace[i].col << '\t' << trace[i].oldCol << endl;
+
string seqAlign = seqs[trace[i].row]->getAligned();
seqAlign = seqAlign.substr(trace[i].col, (trace[i].oldCol-trace[i].col+1));
chimera += seqAlign;
exit(1);
}
}
+
+//***************************************************************************************************************
+
+string Maligner::constructAntiChimericSeq(vector<trace_struct> trace, vector<Sequence*> seqs) {
+ try {
+ string antiChimera = "";
+
+ for (int i = 0; i < trace.size(); i++) {
+// cout << i << '\t' << (trace.size() - i - 1) << '\t' << trace[i].row << '\t' << trace[i].col << '\t' << trace[i].oldCol << endl;
+
+ int oppositeIndex = trace.size() - i - 1;
+
+ string seqAlign = seqs[trace[oppositeIndex].row]->getAligned();
+ seqAlign = seqAlign.substr(trace[i].col, (trace[i].oldCol-trace[i].col+1));
+ antiChimera += seqAlign;
+ }
+
+ return antiChimera;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Maligner", "constructChimericSeq");
+ exit(1);
+ }
+}
+
//***************************************************************************************************************
float Maligner::computePercentID(string queryAlign, string chimera) {
try {
}
}
//***************************************************************************************************************
-vector<Sequence*> Maligner::getBlastSeqs(Sequence* q, int num) {
- try {
- indexes.clear();
- vector<Sequence*> refResults;
- //generate blastdb
- Database* database = new BlastDB(-2.0, -1.0, matchScore, misMatchPenalty);
- for (int i = 0; i < db.size(); i++) { database->addSequence(*db[i]); }
- database->generateDB();
- database->setNumSeqs(db.size());
-
- //get parts of query
- string queryUnAligned = q->getUnaligned();
- string leftQuery = queryUnAligned.substr(0, int(queryUnAligned.length() * 0.33)); //first 1/3 of the sequence
- string rightQuery = queryUnAligned.substr(int(queryUnAligned.length() * 0.66)); //last 1/3 of the sequence
-
- Sequence* queryLeft = new Sequence(q->getName(), leftQuery);
- Sequence* queryRight = new Sequence(q->getName(), rightQuery);
-
- vector<int> tempIndexesRight = database->findClosestMegaBlast(queryRight, num+1);
- vector<int> tempIndexesLeft = database->findClosestMegaBlast(queryLeft, num+1);
-
- //if ((tempIndexesRight.size() != (num+1)) || (tempIndexesLeft.size() != (num+1))) { m->mothurOut("megablast returned " + toString(tempIndexesRight.size()) + " results for the right end, and " + toString(tempIndexesLeft.size()) + " for the left end. Needed " + toString(num+1) + ". Unable to porcess sequence " + q->getName()); m->mothurOutEndLine(); return refResults; }
-
- vector<int> smaller;
- vector<int> larger;
-
- if (tempIndexesRight.size() < tempIndexesLeft.size()) { smaller = tempIndexesRight; larger = tempIndexesLeft; }
- else { smaller = tempIndexesLeft; larger = tempIndexesRight; }
-
- //merge results
- map<int, int> seen;
- map<int, int>::iterator it;
-
- vector<int> mergedResults;
- for (int i = 0; i < smaller.size(); i++) {
- //add left if you havent already
- it = seen.find(smaller[i]);
- if (it == seen.end()) {
- mergedResults.push_back(smaller[i]);
- seen[smaller[i]] = smaller[i];
- }
-
- //add right if you havent already
- it = seen.find(larger[i]);
- if (it == seen.end()) {
- mergedResults.push_back(larger[i]);
- seen[larger[i]] = larger[i];
- }
- }
-
- for (int i = smaller.size(); i < larger.size(); i++) {
- it = seen.find(larger[i]);
- if (it == seen.end()) {
- mergedResults.push_back(larger[i]);
- seen[larger[i]] = larger[i];
- }
- }
-
- if (mergedResults.size() < numWanted) { numWanted = mergedResults.size(); }
-//cout << q->getName() << endl;
- for (int i = 0; i < numWanted; i++) {
-//cout << db[mergedResults[i]]->getName() << endl;
- if (db[mergedResults[i]]->getName() != q->getName()) {
- Sequence* temp = new Sequence(db[mergedResults[i]]->getName(), db[mergedResults[i]]->getAligned());
- refResults.push_back(temp);
- indexes.push_back(mergedResults[i]);
- }
-//cout << mergedResults[i] << endl;
- }
-//cout << endl;
- delete queryRight;
- delete queryLeft;
- delete database;
-
- return refResults;
- }
- catch(exception& e) {
- m->errorOut(e, "Maligner", "getBlastSeqs");
- exit(1);
- }
-}
-//***************************************************************************************************************
-vector<Sequence*> Maligner::getKmerSeqs(Sequence* q, int num) {
- try {
- indexes.clear();
-
- //get parts of query
- string queryUnAligned = q->getUnaligned();
- string leftQuery = queryUnAligned.substr(0, int(queryUnAligned.length() * 0.33)); //first 1/3 of the sequence
- string rightQuery = queryUnAligned.substr(int(queryUnAligned.length() * 0.66)); //last 1/3 of the sequence
-
- Sequence* queryLeft = new Sequence(q->getName(), leftQuery);
- Sequence* queryRight = new Sequence(q->getName(), rightQuery);
-
- vector<int> tempIndexesLeft = databaseLeft->findClosestSequences(queryLeft, numWanted);
- vector<int> tempIndexesRight = databaseRight->findClosestSequences(queryRight, numWanted);
-
- //merge results
- map<int, int> seen;
- map<int, int>::iterator it;
-
- vector<int> mergedResults;
- for (int i = 0; i < tempIndexesLeft.size(); i++) {
- //add left if you havent already
- it = seen.find(tempIndexesLeft[i]);
- if (it == seen.end()) {
- mergedResults.push_back(tempIndexesLeft[i]);
- seen[tempIndexesLeft[i]] = tempIndexesLeft[i];
- }
-
- //add right if you havent already
- it = seen.find(tempIndexesRight[i]);
- if (it == seen.end()) {
- mergedResults.push_back(tempIndexesRight[i]);
- seen[tempIndexesRight[i]] = tempIndexesRight[i];
- }
- }
-
-//cout << q->getName() << endl;
- vector<Sequence*> refResults;
- for (int i = 0; i < numWanted; i++) {
-//cout << db[mergedResults[i]]->getName() << endl;
- Sequence* temp = new Sequence(db[mergedResults[i]]->getName(), db[mergedResults[i]]->getAligned());
- refResults.push_back(temp);
- indexes.push_back(mergedResults[i]);
- }
-//cout << endl;
- delete queryRight;
- delete queryLeft;
-
- return refResults;
- }
- catch(exception& e) {
- m->errorOut(e, "Maligner", "getBlastSeqs");
- exit(1);
- }
-}
-//***************************************************************************************************************
-
projects / mothur.git / blobdiff