OpenStructure: stage/include/ost/io/mmcif

00001 //------------------------------------------------------------------------------
00002 // This file is part of the OpenStructure project <www.openstructure.org>
00003 //
00004 // Copyright (C) 2008-2011 by the OpenStructure authors
00005 //
00006 // This library is free software; you can redistribute it and/or modify it under
00007 // the terms of the GNU Lesser General Public License as published by the Free
00008 // Software Foundation; either version 3.0 of the License, or (at your option)
00009 // any later version.
00010 // This library is distributed in the hope that it will be useful, but WITHOUT
00011 // ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
00012 // FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for more
00013 // details.
00014 //
00015 // You should have received a copy of the GNU Lesser General Public License
00016 // along with this library; if not, write to the Free Software Foundation, Inc.,
00017 // 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
00018 //------------------------------------------------------------------------------
00019 #ifndef OST_MMCIF_READER_HH
00020 #define OST_MMCIF_READER_HH
00021 
00022 #include <map>
00023 
00024 #include <ost/geom/geom.hh>
00025 #include <ost/seq/sequence_list.hh>
00026 #include <ost/mol/residue_handle.hh>
00027 #include <ost/mol/chain_type.hh>
00028 #include <ost/conop/compound_lib.hh>
00029 #include <ost/io/mol/io_profile.hh>
00030 #include <ost/io/io_exception.hh>
00031 #include <ost/io/mol/star_parser.hh>
00032 #include <ost/io/mol/mmcif_info.hh>
00033 
00034 namespace ost { namespace io {
00035 
00059 class DLLEXPORT_OST_IO MMCifReader : public StarParser  {
00060 public:
00066 
00067   MMCifReader(std::istream& stream, mol::EntityHandle& ent_handle,
00068               const IOProfile& profile);
00069 
00075   MMCifReader(const String& filename, mol::EntityHandle& ent_handle,
00076               const IOProfile& profile);
00077 
00080   void Init();
00081 
00083   void ClearState();
00084 
00088   void SetRestrictChains(const String& restrict_chains);
00089 
00095   void SetReadCanonicalSeqRes(bool flag)
00096   {
00097     seqres_can_ = flag;
00098   }
00099 
00100   const String& GetRestrictChains() const
00101   {
00102     return restrict_chains_;
00103   }
00104 
00109   void SetAuthChainID(bool id)
00110   {
00111     auth_chain_id_ = id;
00112   }
00113 
00120   virtual bool OnBeginData(const StringRef& data_name);
00121 
00127   virtual bool OnBeginLoop(const StarLoopDesc& header); // tested
00128 
00133   virtual void OnDataRow(const StarLoopDesc& header, 
00134                          const std::vector<StringRef>& columns);
00135 
00137   virtual void OnEndData();
00138 
00142   seq::SequenceList GetSeqRes() const {
00143     return seqres_;
00144   }
00145 
00149   void SetReadSeqRes(bool flag)
00150   {
00151     read_seqres_ = flag;
00152   }
00153 
00157   bool GetReadSeqRes() const
00158   {
00159     return read_seqres_;
00160   }
00161 
00165   const MMCifInfo& GetInfo() { return info_; }
00166 
00167 protected:
00174   void TryStoreIdx(const int mapping,
00175                    const String& item,
00176                    const StarLoopDesc& header)
00177   {
00178     indices_[mapping] = header.GetIndex(item);
00179     
00180     if (indices_[mapping] == -1) {
00181       throw IOException(this->FormatDiagnostic(STAR_DIAG_ERROR,
00182                                                "No item '" + item +
00183                                                "' found in '" +
00184                                                header.GetCategory()+
00185                                                "' header",
00186                                                this->GetCurrentLinenum()));
00187      }
00188   } // tested
00189 
00195   bool IsValidPDBIdent(const StringRef& pdbid);
00196 
00210   bool ParseAtomIdent(const std::vector<StringRef>& columns,
00211                       String& auth_chain_name,
00212                       String& cif_chain_name,
00213                       StringRef& res_name,
00214                       mol::ResNum& resnum,
00215                       bool& valid_res_num,
00216                       StringRef& atom_name,
00217                       char& alt_loc);
00218 
00222   void ParseAndAddAtom(const std::vector<StringRef>& columns);
00223 
00227   void ParseEntity(const std::vector<StringRef>& columns);
00228 
00232   void ParseEntityPoly(const std::vector<StringRef>& columns);
00233 
00237   void ParseCitation(const std::vector<StringRef>& columns);
00238 
00239   const MMCifInfoStructRefs& GetStructRefs() const { return struct_refs_; }
00248   String ConvertSEQRES(const String& seqres, conop::CompoundLibPtr compound_lib);
00252   void ParseCitationAuthor(const std::vector<StringRef>& columns);
00253   
00255   void ParseStructRef(const std::vector<StringRef>& columns);
00256   
00258   void ParseStructRefSeq(const std::vector<StringRef>& columns);
00259 
00261   void ParseStructRefSeqDif(const std::vector<StringRef>& columns);
00265   void ParseExptl(const std::vector<StringRef>& columns);
00266 
00270   void ParseRefine(const std::vector<StringRef>& columns);
00271 
00275   void ParsePdbxStructAssembly(const std::vector<StringRef>& columns);
00276 
00280   void ParsePdbxStructAssemblyGen(const std::vector<StringRef>& columns);
00281 
00282   std::vector<std::vector<String> > UnPackOperExperession(StringRef expression);
00283 
00284   void StoreExpression(const char* l, const char* s,
00285                        bool& is_range, int lborder,
00286                        std::vector<String>& single_block);
00287 
00288   void StoreRange(const char*& l, const char* s, bool& is_range, int& lborder,
00289                   std::vector<String>& single_block);
00290 
00294   void ParsePdbxStructOperList(const std::vector<StringRef>& columns);
00295 
00299   void ParseDatabasePDBRev(const std::vector<StringRef>& columns);
00300 
00304   void ParseStruct(const std::vector<StringRef>& columns);
00305 
00309   void ParseStructConf(const std::vector<StringRef>& columns);
00310 
00314   void ParseStructSheetRange(const std::vector<StringRef>& columns);
00315 
00319   void ParsePdbxDatabasePdbObsSpr(const std::vector<StringRef>& columns);
00320 
00322   typedef enum {
00323     MMCIF_HELIX,
00324     MMCIF_STRAND,
00325     MMCIF_TURN
00326   } MMCifSecStructElement;
00327 
00331   MMCifSecStructElement DetermineSecStructType(const StringRef& type) const;
00332 
00336   void AssignSecStructure(mol::EntityHandle ent);
00337 
00338 private:
00340   typedef enum {
00341     PDBID_LEN=4,         
00342     MAX_ITEMS_IN_ROW=18, 
00343   } MMCifMagicNos;
00344 
00346   typedef enum {
00347     AUTH_ASYM_ID,      
00348     AS_ID,             
00349     LABEL_ALT_ID,      
00350     LABEL_ASYM_ID,     
00351     LABEL_ATOM_ID,
00352     LABEL_COMP_ID,
00353     LABEL_ENTITY_ID,   
00354     LABEL_SEQ_ID,      
00355     AUTH_SEQ_ID,       
00356     TYPE_SYMBOL,       
00357     CARTN_X,           
00358     CARTN_Y,           
00359     CARTN_Z,           
00360     OCCUPANCY,
00361     B_ISO_OR_EQUIV,
00362     PDBX_PDB_INS_CODE,
00363     GROUP_PDB,         
00364     PDBX_PDB_MODEL_NUM 
00365   } AtomSiteItems;
00366 
00368   typedef enum {
00369     E_ID,              
00370     E_TYPE,            
00371     PDBX_DESCRIPTION   
00372   } EntityItems;
00373 
00375   typedef enum {
00376     ENTITY_ID,                    
00377     EP_TYPE,                      
00378     PDBX_SEQ_ONE_LETTER_CODE,     
00379     PDBX_SEQ_ONE_LETTER_CODE_CAN  
00380   } EntityPolyItems;
00381 
00383   typedef enum {
00384     CITATION_ID,                  
00385     ABSTRACT_ID_CAS,              
00386     BOOK_ID_ISBN,                 
00387     BOOK_TITLE,                   
00388     JOURNAL_ABBREV,               
00389     JOURNAL_VOLUME,               
00390     PAGE_FIRST,                   
00391     PAGE_LAST,                    
00392     PDBX_DATABASE_ID_DOI,         
00393     PDBX_DATABASE_ID_PUBMED,      
00394     YEAR,                         
00395     TITLE                         
00396   } CitationItems;
00397 
00399   typedef enum {
00400     AUTHOR_CITATION_ID,           
00401     AUTHOR_NAME,                  
00402     ORDINAL                       
00403   } CitationAuthorItems;
00404 
00406   typedef enum {
00407     EXPTL_ENTRY_ID,               
00408     METHOD                        
00409   } ExptlItems;
00410 
00412   typedef enum {
00413     REFINE_ENTRY_ID,              
00414     LS_D_RES_HIGH,                
00415     LS_D_RES_LOW
00416   } RefineItems;
00417 
00419   typedef enum {
00420     PSA_DETAILS,                  
00421     PSA_ID,                       
00422     METHOD_DETAILS                
00423   } PdbxStructAssemblyItems;
00424 
00425   // \enum items of the struct_ref category
00426   typedef enum {
00427     SR_ENTITY_ID,
00428     SR_ID,
00429     SR_DB_CODE,
00430     SR_DB_NAME,
00431     SR_DB_ACCESS
00432   } StructRefItems;
00433   
00435   typedef enum {
00436     SRS_ALIGN_ID,
00437     SRS_STRUCT_REF_ID,
00438     SRS_PDBX_STRAND_ID,
00439     SRS_DB_ALIGN_BEG,
00440     SRS_DB_ALIGN_END,
00441     SRS_ENT_ALIGN_BEG,
00442     SRS_ENT_ALIGN_END
00443   } StructRefSeqItems;
00444   
00446   typedef enum {
00447     SRSD_ALIGN_ID,
00448     SRSD_SEQ_RNUM,
00449     SRSD_DB_RNUM,
00450     SRSD_DETAILS
00451   } StructRefSeqDifItems;
00452 
00454   typedef enum {
00455     ASSEMBLY_ID,                  
00456     ASYM_ID_LIST,                 
00457     OPER_EXPRESSION               
00458   } PdbxStructAssemblyGenItems;
00459 
00461   typedef enum {
00462     PSOL_ID,    
00463     PSOL_TYPE,  
00464     VECTOR_1,   
00465     VECTOR_2,   
00466     VECTOR_3,   
00467     MATRIX_1_1, 
00468     MATRIX_1_2, 
00469     MATRIX_1_3, 
00470     MATRIX_2_1, 
00471     MATRIX_2_2, 
00472     MATRIX_2_3, 
00473     MATRIX_3_1, 
00474     MATRIX_3_2, 
00475     MATRIX_3_3  
00476   } PdbxStructOperListItems;
00477 
00479   typedef enum {
00480      STRUCT_ENTRY_ID,     
00481      PDBX_CASP_FLAG,      
00482      PDBX_DESCRIPTOR,     
00483      PDBX_FORMULA_WEIGHT, 
00484      PDBX_FORMULA_WEIGHT_METHOD, 
00485      PDBX_MODEL_DETAILS,  
00486      PDBX_MODEL_TYPE_DETAILS, 
00487      STRUCT_TITLE         
00488   } StructItems;
00489 
00491   typedef enum {
00492     SC_BEG_AUTH_ASYM_ID,  
00493     SC_BEG_LABEL_ASYM_ID, 
00494     SC_BEG_LABEL_COMP_ID, 
00495     SC_BEG_LABEL_SEQ_ID,  
00496     SC_CONF_TYPE_ID,      
00497     SC_END_AUTH_ASYM_ID,  
00498     SC_END_LABEL_ASYM_ID, 
00499     SC_END_LABEL_COMP_ID, 
00500     SC_END_LABEL_SEQ_ID,  
00501     SC_ID,                
00502   } StructConfItems;
00503 
00505   typedef enum {
00506     SSR_BEG_LABEL_ASYM_ID,     
00507     SSR_BEG_LABEL_COMP_ID,     
00508     SSR_BEG_LABEL_SEQ_ID,    
00509     SSR_END_LABEL_ASYM_ID,     
00510     SSR_END_LABEL_COMP_ID,     
00511     SSR_END_LABEL_SEQ_ID,      
00512     SSR_SHEET_ID,              
00513     SSR_ID,                    
00514     SSR_BEG_AUTH_ASYM_ID,      
00515     SSR_END_AUTH_ASYM_ID,      
00516     } StructSheetRangeItems;
00517 
00519   typedef enum {
00520     DATE,           
00521     PDPOS_ID,       
00522     PDB_ID,         
00523     REPLACE_PDB_ID, 
00524   } PdbxDatabasePDBObsSpr;
00525 
00527   typedef enum {
00528     DPI_NUM,           
00529     DPI_DATE,          
00530     DPI_DATE_ORIGINAL, 
00531     DPI_STATUS,        
00532   } DatabasePDBRevItems;
00533 
00535   typedef enum {
00536     ATOM_SITE,
00537     ENTITY,
00538     ENTITY_POLY,
00539     CITATION,
00540     CITATION_AUTHOR,
00541     EXPTL,
00542     REFINE,
00543     PDBX_STRUCT_ASSEMBLY,
00544     PDBX_STRUCT_ASSEMBLY_GEN,
00545     PDBX_STRUCT_OPER_LIST,
00546     STRUCT,
00547     STRUCT_CONF,
00548     STRUCT_SHEET_RANGE,
00549     PDBX_DATABASE_PDB_OBS_SPR,
00550     STRUCT_REF,
00551     STRUCT_REF_SEQ,
00552     STRUCT_REF_SEQ_DIF,
00553     DATABASE_PDB_REV,
00554     DONT_KNOW
00555   } MMCifCategory;
00556 
00558   typedef struct {
00559     mol::ChainType type; 
00560     String details;      
00561     String seqres;       
00562   } MMCifEntityDesc;
00563   typedef std::map<String, MMCifEntityDesc> MMCifEntityDescMap;
00564 
00566   typedef struct {
00567     String biounit_id;                              
00568     std::vector<String> chains;                     
00569 
00570     std::vector<std::vector<String> > operations;   
00571 
00572   } MMCifBioUAssembly;
00573   typedef std::vector<MMCifBioUAssembly> MMCifBioUAssemblyVector;
00574 
00575   typedef std::map<String, std::pair<std::vector<int>, std::vector<String> > >
00576     MMCifCitationAuthorMap;
00577 
00579   typedef struct {
00580     mol::ResNum start;
00581     mol::ResNum end;
00582     String chain_name;
00583   } MMCifHSEntry;
00584   typedef std::vector<MMCifHSEntry> MMCifHSVector;
00585 
00587   typedef struct {
00588     String details;
00589     String method_details;
00590   } MMCifPSAEntry;
00591   typedef std::map<String, MMCifPSAEntry> MMCifPSAMap;
00592 
00593   // members
00594   MMCifCategory category_;
00595   int category_counts_[DONT_KNOW+1]; 
00596   int indices_[MAX_ITEMS_IN_ROW]; 
00597   const IOProfile& profile_;
00598   mol::EntityHandle& ent_handle_;
00599   String restrict_chains_;
00600   bool auth_chain_id_;       
00601   bool seqres_can_;          
00602   mol::ChainHandle curr_chain_;
00603   mol::ResidueHandle curr_residue_;
00604   int chain_count_;
00605   int residue_count_;
00606   int atom_count_;
00607   bool warned_name_mismatch_;
00608   bool warned_rule_based_;
00609   String subst_res_id_; 
00610   bool has_model_;      
00611   int curr_model_;      
00612   std::vector<std::pair<mol::ChainHandle, String> > chain_id_pairs_;
00614   MMCifEntityDescMap entity_desc_map_; 
00615   seq::SequenceList seqres_;
00616   bool read_seqres_;
00617   MMCifInfo info_;      
00618   MMCifCitationAuthorMap authors_map_;
00619   MMCifBioUAssemblyVector bu_assemblies_;
00620   MMCifPSAMap bu_origin_map_; 
00621   MMCifHSVector helix_list_; 
00622   MMCifHSVector strand_list_; 
00623   MMCifInfoStructRefs struct_refs_;
00624 };
00625 
00626 }}
00627 
00628 #endif
stage/include/ost/io/mmcif_reader.hh
