#include "chimera.h"
#include "dist.h"
+#include "decalc.h"
+/***********************************************************/
//This class was created using the algorythms described in the
// "At Least 1 in 20 16S rRNA Sequence Records Currently Held in the Public Repositories is Estimated To Contain Substantial Anomalies" paper
//by Kevin E. Ashelford 1, Nadia A. Chuzhanova 3, John C. Fry 1, Antonia J. Jones 2 and Andrew J. Weightman 1.
class Pintail : public Chimera {
public:
- Pintail(string, string);
+ Pintail(string, string, bool, int, string, string, string, int, int, string); //fastafile, templatefile, filter, processors, mask, conservation, quantile, window, increment, outputDir)
~Pintail();
- void getChimeras();
- void print(ostream&);
+ int getChimeras(Sequence*);
+ Sequence print(ostream&, ostream&);
- void setCons(string c) { consfile = c; }
+ void setCons(string c) { consfile = c; }
+ void setQuantiles(string q) { quanfile = q; }
+ #ifdef USE_MPI
+ Sequence print(MPI_File&, MPI_File&);
+ #endif
private:
- struct linePair {
- int start;
- int end;
- linePair(int i, int j) : start(i), end(j) {}
- };
-
Dist* distcalculator;
- int iters;
- string fastafile, templateFile, consfile;
- vector<linePair*> lines;
- vector<Sequence*> querySeqs;
- vector<Sequence*> templateSeqs;
+ DeCalculator* decalc;
+ int iters, window, increment, processors;
+ string fastafile, quanfile, consfile;
- vector<Sequence> bestfit; //bestfit[0] matches queryseqs[0]...
+ vector<linePair*> templateLines;
+ Sequence* querySeq;
+
+ Sequence* bestfit; //closest match to query in template
- vector< vector<float> > obsDistance; //obsDistance[0] is the vector of observed distances for queryseqs[0]...
- vector< vector<float> > expectedDistance; //expectedDistance[0] is the vector of expected distances for queryseqs[0]...
- vector<float> deviation; //deviation[0] is the percentage of mismatched pairs over the whole seq between querySeqs[0] and its best match.
- vector< vector<int> > windows; // windows[0] is a vector containing the starting spot in queryseqs[0] aligned sequence for each window.
+ vector<float> obsDistance; //obsDistance is the vector of observed distances for query
+ vector<float> expectedDistance; //expectedDistance is the vector of expected distances for query
+ float deviation; //deviation is the percentage of mismatched pairs over the whole seq between query and its best match.
+ vector<int> windowsForeachQuery; // windowsForeachQuery is a vector containing the starting spot in query aligned sequence for each window.
//this is needed so you can move by bases and not just spots in the alignment
- vector< map<int, int> > trim; //trim[0] is the start and end position of trimmed querySeqs[0]. Used to find the variability over each sequence window.
- vector<int> windowSizes; //windowSizes[0] = window size of querySeqs[0]
+ int windowSizes; //windowSizes = window size of query
+ vector<int> windowSizesTemplate; //windowSizesTemplate[0] = window size of templateSeqs[0]
- vector< vector<float> > Qav; //Qav[0] is the vector of average variablility for queryseqs[0]...
- vector<float> seqCoef; //seqCoef[0] is the coeff for queryseqs[0]...
- vector<float> DE; //DE[0] is the deviaation for queryseqs[0]...
+ map<int, int> trimmed; //trimmed = start and stop of trimmed sequences for query
+ map<int, int>::iterator it;
+
+ vector<float> Qav; //Qav is the vector of average variablility for query
+ float seqCoef; //seqCoef is the coeff for query
+ float DE; //DE is the deviaation for query
vector<float> probabilityProfile;
+ vector< vector<float> > quantiles; //quantiles[0] is the vector of deviations with ceiling score of 1, quantiles[1] is the vector of deviations with ceiling score of 2...
+ vector< vector<float> > quantilesMembers; //quantiles[0] is the vector of deviations with ceiling score of 1, quantiles[1] is the vector of deviations with ceiling score of 2...
+ set<int> h;
+ string mergedFilterString;
- vector<Sequence*> readSeqs(string);
- void trimSeqs(Sequence*, Sequence&, int);
+ vector< vector<float> > readQuantiles();
vector<float> readFreq();
- vector< vector<float> > findQav(int, int);
- vector<float> calcFreq(vector<Sequence*>);
- vector<float> getCoef(int, int);
-
- vector<Sequence> findPairs(int, int);
- vector< vector<int> > findWindows(int, int);
- vector< vector<float> > calcObserved(int, int);
- vector< vector<float> > calcExpected(int, int);
- vector<float> calcDE(int, int);
- vector<float> calcDist(int, int);
-
- void createProcessesSpots();
- void createProcesses();
-
-
+ Sequence* findPairs(Sequence*);
+
+ void createProcessesQuan();
+ int doPrep();
+ void printQuanFile(string, string);
};
projects / mothur.git / blobdiff