kzblast.h

Go to the documentation of this file.
00001 #ifndef KZBLAST_H
00002 #define KZBLAST_H
00003 
00004 #include "linktrie.h"
00005 #include "alignseg.h"
00006 #include <list>
00007 
00008 using namespace std;
00009 
00010 struct BlastParameter {
00011    BlastParameter() : wordSize(3), neighborThreshold(wordSize*3.3),
00012       neighborFraction(0.7), wordjoin(4), extDropoff(-1),
00013      extCutoff(11), gapOpen(-11), gapExtend(-1), gapAlignDropOff(12) { }
00014 
00015    int wordSize;
00016    float neighborThreshold; // neighborhood word score threshold
00017    float neighborFraction;  // > fraction of self score will be saved
00018    int wordjoin;  // word distance to join 
00019    int extDropoff; // ungapped extension dropoff
00020    int extCutoff;  // threshold for hit extension: blastp 11, blastn 0, blastx 12, tblastn 13, tblastx 13, megablast 0
00021    int gapOpen;
00022    int gapExtend;
00023    int gapAlignDropOff;
00024 };
00025 
00034 class Lale {
00035    public:
00036       Lale() : qseq(0), mtx("blosum62.50"), param(), fsa(new ltrie(4)), maxss() { }
00037       Lale(int ws, const BlastParameter& bp) 
00038          : mtx("blosum62.50"), param(bp), fsa(new ltrie(ws)), maxss() { }
00039       void readQuery(const string &file);
00040       void buildfsa() { fsa->build(qseq->toString(), mtx, param.neighborThreshold, param.neighborFraction); }
00041 
00044       void showfsa(ostream &ous) const { fsa->showLeafs(ous); }
00048       list<Alignseg>& findMatch(const bioseq &s);
00059       int extendhit(Alignseg &as, const int* tc, const int tclen);
00060       void showseg_debug(const Alignseg &as, const bioseq& ss) const;
00061       void showmaxss_debug(const bioseq* tseq) const;
00062       // to see the Alignseg is the complete alignment or not
00067       bool isComplete(const Alignseg& as, const bioseq &ss);
00068       void setMatrix(const Matrix& mt) { mtx=mt; }
00069       const Matrix* getMatrix() const { return &mtx; }
00070       const bioseq* getQuery() const { return qseq; }
00071 
00072    private:
00073       void mergeNeighbor(list<Alignseg> &segs, const int* qsc, const int* tsc);
00074       bioseq *qseq;
00075       Matrix mtx;
00076       BlastParameter param;
00077       list<Alignseg> maxss; // maximal scoring segments
00078       ltrie* fsa;
00079 };
00080 
00081 #endif