gnSequence.h Source File

00001 
00002 // File:            gnSequence.h
00003 // Purpose:         Sequence class
00004 // Description:     Provides a high level sequence interface to all types of
00005 //                                      sequence data.
00006 // Changes:        
00007 // Version:         libGenome 0.1.0 
00008 // Author:          Aaron Darling 
00009 // Last Edited:     April 15, 2001, 10:34:50pm 
00010 // Modified by:     
00011 // Copyright:       (c) Aaron Darling 
00012 // Licenses:        Proprietary 
00014 #ifndef _gnSequence_h_
00015 #define _gnSequence_h_
00016 
00017 #include "gn/gnDefs.h"
00018 
00019 #include <string>
00020 #include <iostream>
00021 #include <list>
00022 #include "gn/gnBaseSource.h"
00023 #include "gn/gnGenomeSpec.h"
00024 #include "gn/gnSourceSpec.h"
00025 #include "gn/gnStringSpec.h"
00026 #include "gn/gnBaseFilter.h"
00027 #include "gn/gnCompare.h"
00028 
00036 class GNDLLEXPORT gnSequence : public gnClone
00037 {
00038 public:
00042         gnSequence();
00047         gnSequence( const gnSeqC* seq );
00052         gnSequence( const string& str );
00057         gnSequence( const gnGenomeSpec& gngs );
00062         gnSequence( const gnFragmentSpec& gnfs );
00068         gnSequence( const gnContigSpec& gncs );
00074         gnSequence( gnSeqC *bases, const gnSeqI length);
00079         gnSequence( const gnSequence& seq);
00083         ~gnSequence();
00084 
00088         gnSequence& operator=( const gnSequence& seq);
00089 
00090         gnSequence* Clone() const;
00091 
00096         gnSeqI contigListSize() const;
00101         gnSeqI contigListLength() const;
00108         uint32 contigIndexByBase( const gnSeqI baseI) const;
00115         gnSequence contig( const uint32 contigI) const;
00122         gnSequence contigByBase( const gnSeqI baseI) const;
00129         virtual gnSeqI contigStart( const uint32 contigI) const;
00136         virtual gnSeqI contigLength( const uint32 contigI) const;
00144         virtual uint32 contigIndexByName( string& contigName) const;
00151         virtual string contigName( const uint32 contigI) const;
00159         virtual gnSequence contigByName( string& contigName) const;
00166         virtual void merge(const gnSeqI startI, const gnSeqI endI);
00173         virtual void mergeContigs(const uint32 startC, const uint32 endC);
00179         virtual void splitContig(const gnSeqI splitI, const uint32 contigI=ALL_CONTIGS);
00180 
00181         virtual void setContigName( const uint32 contigI, const string& contig_name);
00182 
00188         virtual uint32 getFeatureListLength() const;
00195         virtual gnBaseFeature* getFeature(const uint32 featureI) const;
00202         virtual void getContainedFeatures(const gnLocation& lt, vector<gnBaseFeature*>& feature_vector, vector<uint32>& index_vector) const;
00209         virtual void getIntersectingFeatures(const gnLocation& lt, vector<gnBaseFeature*>& feature_vector, vector<uint32>& index_vector) const;
00215         virtual uint32 addFeature(gnBaseFeature* feature);
00221         virtual void removeFeature(const uint32 featureI);
00227         virtual void getBrokenFeatures(const gnLocation& lt, vector<gnBaseFeature*>& feature_vector) const;
00234         virtual uint32 getHeaderListLength(const uint32 contigI) const;
00243         virtual gnBaseHeader* getHeader(const uint32 contigI, const uint32 headerI) const;
00251         virtual void addHeader(const uint32 contigI, gnBaseHeader* header, const uint32 headerI);
00258         virtual void removeHeader(const uint32 contigI, const uint32 headerI);
00265         virtual void setReverseComplement( const boolean revComp, const uint32 contigI=ALL_CONTIGS);
00272         virtual boolean isReverseComplement( const uint32 contigI=ALL_CONTIGS );
00277         virtual boolean isCircular() const;
00284         virtual void setCircular( const boolean value );
00285         
00292         virtual void globalToLocal(uint32& contigI, gnSeqI& baseI) const;
00293 
00301         virtual void localToGlobal(const uint32 contigI, gnSeqI& baseI) const;
00309         virtual void globalToSource(uint32& contigI, gnSeqI& baseI) const;
00319         virtual void localToSource(uint32& contigI, gnSeqI& baseI) const;
00328         virtual bool LoadSource(const string sourcename);
00329 
00334         virtual void setFilter(const gnBaseFilter* filt);
00335 
00341         virtual void setFilterList(list<const gnBaseFilter*>& filt_list);
00342 
00346         virtual list<const gnBaseFilter*> getFilterList() const;
00347 
00352         virtual void assign(gnSequence& seq);
00353 
00354 // Comparison Operators
00358         void operator=(gnSequence& seq);
00359         boolean operator==(const gnSequence& seq) const;
00360         boolean operator!=(const gnSequence& seq) const;
00361         boolean operator<(const gnSequence& seq) const;
00362         boolean operator<=(const gnSequence& seq) const;
00363         boolean operator>(const gnSequence& seq) const;
00364         boolean operator>=(const gnSequence& seq) const;
00368         gnSequence& operator+=(const gnSequence& seq);
00375         virtual int compare(const gnSequence& seq) const;
00376         virtual int compare(const string& str) const;
00377 // Append and Insert Operators
00382         virtual void append( const gnSequence& seq);
00390         virtual void insert( const gnSeqI offset, const gnSeqC *bases, const gnSeqI length);
00397         virtual void insert( const gnSeqI offset, const gnSequence& seq);
00404         virtual void insert( const gnSeqI offset, const gnGenomeSpec& gnbs);
00405 // Concatenation Operators
00409         gnSequence const operator+(const gnSequence& seq) const;
00410 // Substrings and base deletion
00417         gnSequence subseq(const gnSeqI offset, const gnSeqI length) const;
00423         virtual void erase( const gnSeqI offset=0, const gnSeqI length=GNSEQI_END );
00424 // IO operators
00428         friend std::istream& operator>>(std::istream& is, gnSequence& gns);     //read from source.
00432         friend std::ostream& operator<<(std::ostream& os, const gnSequence& gns); //write to source.
00433 // Size functions
00438         virtual gnSeqI length() const;
00443         virtual gnSeqI size() const;
00444 // Raw Sequence Access
00451         virtual string ToString( const gnSeqI length=GNSEQI_END, const gnSeqI offset=1 ) const;
00459         virtual boolean ToString( string& str, const gnSeqI length=GNSEQI_END, const gnSeqI offset=1 ) const;
00469         virtual boolean ToArray( gnSeqC* pSeqC, gnSeqI length, const gnSeqI offset=1 ) const;
00475         virtual gnSeqC GetSeqC( const gnSeqI offset ) const; // return gnSeqC illegal char.
00480         gnSeqC operator[]( const gnSeqI offset ) const;
00481 
00486         virtual gnGenomeSpec* GetSpec() const;
00491         virtual void SetSpec(gnGenomeSpec* s);
00492         
00501         virtual gnSeqI find(const gnSequence& search, const gnSeqI offset=0) const;
00502         
00503 private:
00504         gnGenomeSpec *spec;
00505         list<const gnBaseFilter*> filter_list;
00506         const gnCompare* comparator;
00507 }; // class gnSequence
00508 
00509 
00510 GNDLLEXPORT
00511 std::istream& operator>>(std::istream& is, gnSequence& gns);    //read from source.
00512 GNDLLEXPORT
00513 std::ostream& operator<<(std::ostream& os, const gnSequence& gns); //write to source.
00514 
00515 // Assignment Operators
00516 inline
00517 void gnSequence::operator=(gnSequence& seq){
00518         spec = seq.spec->Clone();
00519 }
00520 inline
00521 void gnSequence::assign(gnSequence& seq){
00522         spec = seq.spec->Clone();
00523 }
00524 inline
00525 boolean gnSequence::operator==(const gnSequence& seq) const{
00526         return (compare(seq)==0);
00527 }
00528 inline
00529 boolean gnSequence::operator!=(const gnSequence& seq) const{
00530         return (compare(seq)!=0);
00531 }
00532 inline
00533 boolean gnSequence::operator<(const gnSequence& seq) const{
00534         return (compare(seq)<0);
00535 }
00536 inline
00537 boolean gnSequence::operator<=(const gnSequence& seq) const{
00538         return (compare(seq)<=0);
00539 }
00540 inline
00541 boolean gnSequence::operator>(const gnSequence& seq) const{
00542         return (compare(seq)>0);
00543 }
00544 inline
00545 boolean gnSequence::operator>=(const gnSequence& seq) const{
00546         return (compare(seq)>=0);
00547 }
00548 // Append and Insert Operators
00549 inline
00550 gnSequence& gnSequence::operator+=(const gnSequence& seq){
00551         insert(GNSEQI_END, *seq.spec);
00552         return *this;
00553 }
00554 inline
00555 void gnSequence::append( const gnSequence& seq){
00556         insert(GNSEQI_END, *seq.spec);
00557 }
00558 //Feature functions
00559 inline
00560 uint32 gnSequence::getFeatureListLength() const{
00561         return spec->GetFeatureListLength();
00562 }
00563 inline
00564 gnBaseFeature* gnSequence::getFeature(const uint32 featureI) const{
00565         return spec->GetFeature(featureI);
00566 }
00567 inline
00568 uint32 gnSequence::addFeature(gnBaseFeature* feature){
00569         return spec->AddFeature(feature);
00570 }
00571 inline
00572 void gnSequence::removeFeature(const uint32 featureI){
00573         spec->RemoveFeature(featureI);
00574 }
00575 inline
00576 void gnSequence::getContainedFeatures(const gnLocation& lt, vector<gnBaseFeature*>& feature_vector, vector<uint32>& index_vector) const{
00577         spec->GetContainedFeatures(lt, feature_vector, index_vector);
00578 }
00579 inline
00580 void gnSequence::getIntersectingFeatures(const gnLocation& lt, vector<gnBaseFeature*>& feature_vector, vector<uint32>& index_vector) const{
00581         spec->GetIntersectingFeatures(lt, feature_vector, index_vector);
00582 }
00583 inline
00584 void gnSequence::getBrokenFeatures(const gnLocation& lt, vector<gnBaseFeature*>& feature_vector) const{
00585         spec->GetBrokenFeatures(lt, feature_vector);
00586 }
00587 
00588 inline
00589 boolean gnSequence::isCircular() const{
00590         return spec->IsCircular();
00591 }
00592 
00593 inline
00594 void gnSequence::setCircular( const boolean value ){
00595         spec->SetCircular(value);
00596 }
00597 inline
00598 void gnSequence::insert( const gnSeqI offset, const gnSequence& seq){
00599         insert(offset, *seq.spec);
00600 }
00601 
00602 // Size functions
00603 inline
00604 gnSeqI gnSequence::length() const{
00605         return spec->GetLength();
00606 }
00607 inline
00608 gnSeqI gnSequence::size() const{
00609         return spec->GetLength();
00610 }
00611 inline
00612 gnGenomeSpec* gnSequence::GetSpec() const{
00613         return spec;
00614 }
00615 inline
00616 void gnSequence::SetSpec(gnGenomeSpec* s){
00617         if(spec != NULL)
00618                 delete spec;
00619         spec = s;
00620 }
00621 
00622 inline
00623 void gnSequence::setFilter(const gnBaseFilter* filt){
00624         filter_list.clear();
00625         if(filt != NULL)
00626                 filter_list.push_back(filt);
00627 }
00628 inline
00629 void gnSequence::setFilterList(list<const gnBaseFilter*>& filt_list){
00630         filter_list = filt_list;
00631 }
00632 inline
00633 list<const gnBaseFilter*> gnSequence::getFilterList() const{
00634         return filter_list;
00635 }
00636 
00637 
00638 #endif
00639         // _gnSequence_h_