feature.h

Go to the documentation of this file.

#ifndef FEATURE_H
#define FEATURE_H
//#include <map>
#include <vector>
#include <string>
#include "seqinfo.h"
#include "dseq.h"
#include "gberr.h"
#include "strformat.h"
class featerr
{ //still use this class, save changes to codes
 public:
  featerr();
        featerr &pick(const string &m);
        featerr &pick(const char *m);
        void print(ostream &ous) const;

 private:
        string info;
};

struct qualif {
        char qkey[20];   //qualifier key
        string qvalue;   // qualifier value

        qualif();
        qualif(const qualif &qu);
        qualif& operator=(const qualif &qu);
        bool keyis(const char* kk) const {return !strcmp(qkey, kk);}
};


class feature
{
 public:
        feature();
        feature(const feature &fe);
        ~feature();
        feature &operator=(const feature &fe);
 
        //
        /* read one feature
         * return 0 for normal,
         * 1 for location parsing error
         * 2 for location format error
         *
         * calls readloc()
         */
        int read(istream &ins, char nxt[]); 

        int outSeg(ostream &ous) const; 

        ostream& outRange(ostream &ous) const;  
        ostream& outRange(ostream &ous, const char ed) const {
                outRange(ous); ous << ed; return ous;
        }  
        ostream& endInfo(ostream &ous, const bool emark=true) const; 

        void output(ostream &seq, ostream& sub, seqinfo &sinf) const;

  /* write out this feature in text format to bd.  No change to obj
        * Mainly useful for debuging.
         */
        void dumpFeat(ostream &bd) const; 

        void writeCDS(ostream& seq, ostream& sub, seqinfo &sinf) const;
        //if cDNA is the source, there is no need for subsequence
        void writemRNA(ostream &seq, ostream& sub, seqinfo &sinf) const;
        void writeSource(ostream &seqout, ostream &subsout, seqinfo &sinf) const;
        void writeGene(ostream &seq, ostream &sub, seqinfo &sinf) const;
        void writeRNA(ostream &sout, ostream &subout, seqinfo &sinf) const;
        void writerpt(ostream &seq, ostream &sub, seqinfo &sinf) const;
        void writeExon(ostream &ous) const;
        void writeToken(ostream &seq, ostream &sub, seqinfo &sinf, 
                                        const dseq &nt) const;

        void writeAllele(ostream &seq, ostream &sub, seqinfo &sinf) const;

        /* write joint Segmented sequence; for mRNA and CDS only */
        void wjoinseg(ostream &sub, seqinfo &sinf) const;

        void writeUTR(ostream &seq, ostream &sub, seqinfo &sinf) const;

        void wFeature(ostream &ous, const seqinfo &sinf) const;         

        //
        bool isAbnormal() const;

        bool complement() const;
        bool newfeat() const {return (findex() == -1);}
        //bool sameloc(const feature &fe);
        /* the current feature is ff or not */
        bool is(const char *ff) const {return !strcmp(curFeat, ff);}

        bool isjoinseg() const;

        /* contains more than one regions join(6..2498,2542..2739 */
        bool locIsJoin() const { return n>0; }

        const char *feat() const { return curFeat; } 

        /*/to implement (23.45)..(204.221) notion in gb format
        we used the following notion
        (begin.vb)..(ve.end) to represent one_of notion
        the default number for vb and ve is 0, which indicates that 
        there is no one_of notion.  
        (begin.end) is indicated by vr. Default value of vr is 0; 1 indicates
        vaguer.  begin^end is simply represented by begin and end.  No flag is set.
        However, this info is not outputed into the file.
        */

        int findex() const; 
        //returns the feature table index of current feature text
        
        bool findQualif(const char* qk, string &vv) const;

        string qvmap(const char *k) const; 

        int qfind(const char *kq) const;

        friend const string orgtag(const seqinfo &sinf);
        //makes a unique short string from full organism name
        //for identifying gene names from different organisms
        //static map<string, string> qual2ace;
        //static void loadQual2ace();

 private:
        int nq;     // length of qv = number of valued qualifiers
        qualif *qv; //qualifier array, may contain duplicated qualifiers
        vector<string> dbxref; //one or more
        vector<string> qnv; //qualifiers withou values
        //string bFeat;  //for all qualifiers from bad feature
        int begin;
        int vb; 
        int end;
        int ve;
        bool noLeft, noRight;
        string replseq; //store allele sequence 
        
        int *loc; //the actual interger data for location info
        int n; //number of location elements. Only set >0 if join
        char *locstr; //location string, for exception handling
        int l, maxl; //l=number of char and maxl=maximium allocated char for locstr
        char curFeat[16]; //current feature text 
        
        void copy(const feature &lo); //helper function

        int readloc(istream &ins, char ln[]);

        static const char *ft[64];
        static const char *qt[74];

        int qindex(const char qual[]) const; 
        //returns the qualifier index of this qualifier in qt table
        //both functions return  -1 to indicate that entry does not exist
        bool mktitle(string &tt) const;

        /* make a key and put result in sk; return status of key making */
        int subkey(string &sk) const;  
        //makes subkey from information in feature and set sk as the key
        //if nothing available, return 0; 1 for finding a key; 2 for 
        //finding protein_id as key necessary for joint objects
        //such as mRNA, CDS etc.

        bool tokey(string &tk) const;  
        
        void rptdump(ostream &ous, const seqinfo &sinf) const;
        void RNAdump(ostream &ous, const seqinfo &sinf) const;
        /* tlnIdx is the index number of translation in the qualifier 
         * vector.  -1 indicates not present */
        void subCDS(ostream &seq, ostream &sub, const string &cdsKey, const string &title, int tlnIdx, const seqinfo &sqinf) const;
};
#endif