OpenStructure: stage/include/ost/io/mol/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 ParsePdbxAuditRevisionHistory(const std::vector<StringRef>& columns);
00305 
00309   void ParsePdbxAuditRevisionDetails(const std::vector<StringRef>& columns);
00310 
00314   void ParsePdbxDatabaseStatus(const std::vector<StringRef>& columns);
00315 
00319   void ParseStruct(const std::vector<StringRef>& columns);
00320 
00324   void ParseStructConf(const std::vector<StringRef>& columns);
00325 
00329   void ParseStructSheetRange(const std::vector<StringRef>& columns);
00330 
00334   void ParsePdbxDatabasePdbObsSpr(const std::vector<StringRef>& columns);
00335 
00337   typedef enum {
00338     MMCIF_HELIX,
00339     MMCIF_STRAND,
00340     MMCIF_TURN
00341   } MMCifSecStructElement;
00342 
00346   MMCifSecStructElement DetermineSecStructType(const StringRef& type) const;
00347 
00351   void AssignSecStructure(mol::EntityHandle ent);
00352 
00353 private:
00355   typedef enum {
00356     PDBID_LEN=4,         
00357     MAX_ITEMS_IN_ROW=18, 
00358   } MMCifMagicNos;
00359 
00361   typedef enum {
00362     AUTH_ASYM_ID,      
00363     AS_ID,             
00364     LABEL_ALT_ID,      
00365     LABEL_ASYM_ID,     
00366     LABEL_ATOM_ID,
00367     LABEL_COMP_ID,
00368     LABEL_ENTITY_ID,   
00369     LABEL_SEQ_ID,      
00370     AUTH_SEQ_ID,       
00371     TYPE_SYMBOL,       
00372     CARTN_X,           
00373     CARTN_Y,           
00374     CARTN_Z,           
00375     OCCUPANCY,
00376     B_ISO_OR_EQUIV,
00377     PDBX_PDB_INS_CODE,
00378     GROUP_PDB,         
00379     PDBX_PDB_MODEL_NUM 
00380   } AtomSiteItems;
00381 
00383   typedef enum {
00384     E_ID,              
00385     E_TYPE,            
00386     PDBX_DESCRIPTION   
00387   } EntityItems;
00388 
00390   typedef enum {
00391     ENTITY_ID,                    
00392     EP_TYPE,                      
00393     PDBX_SEQ_ONE_LETTER_CODE,     
00394     PDBX_SEQ_ONE_LETTER_CODE_CAN  
00395   } EntityPolyItems;
00396 
00398   typedef enum {
00399     CITATION_ID,                  
00400     ABSTRACT_ID_CAS,              
00401     BOOK_ID_ISBN,                 
00402     BOOK_TITLE,                   
00403     JOURNAL_ABBREV,               
00404     JOURNAL_VOLUME,               
00405     PAGE_FIRST,                   
00406     PAGE_LAST,                    
00407     PDBX_DATABASE_ID_DOI,         
00408     PDBX_DATABASE_ID_PUBMED,      
00409     YEAR,                         
00410     TITLE                         
00411   } CitationItems;
00412 
00414   typedef enum {
00415     AUTHOR_CITATION_ID,           
00416     AUTHOR_NAME,                  
00417     ORDINAL                       
00418   } CitationAuthorItems;
00419 
00421   typedef enum {
00422     EXPTL_ENTRY_ID,               
00423     METHOD                        
00424   } ExptlItems;
00425 
00427   typedef enum {
00428     REFINE_ENTRY_ID,              
00429     LS_D_RES_HIGH,                
00430     LS_D_RES_LOW,
00431     LS_R_FACTOR_R_WORK,           
00432     LS_R_FACTOR_R_FREE            
00433   } RefineItems;
00434 
00436   typedef enum {
00437     PSA_DETAILS,                  
00438     PSA_ID,                       
00439     METHOD_DETAILS                
00440   } PdbxStructAssemblyItems;
00441 
00442   // \enum items of the struct_ref category
00443   typedef enum {
00444     SR_ENTITY_ID,
00445     SR_ID,
00446     SR_DB_CODE,
00447     SR_DB_NAME,
00448     SR_DB_ACCESS
00449   } StructRefItems;
00450   
00452   typedef enum {
00453     SRS_ALIGN_ID,
00454     SRS_STRUCT_REF_ID,
00455     SRS_PDBX_STRAND_ID,
00456     SRS_DB_ALIGN_BEG,
00457     SRS_DB_ALIGN_END,
00458     SRS_ENT_ALIGN_BEG,
00459     SRS_ENT_ALIGN_END
00460   } StructRefSeqItems;
00461   
00463   typedef enum {
00464     SRSD_ALIGN_ID,
00465     SRSD_SEQ_RNUM,
00466     SRSD_DB_RNUM,
00467     SRSD_DETAILS
00468   } StructRefSeqDifItems;
00469 
00471   typedef enum {
00472     ASSEMBLY_ID,                  
00473     ASYM_ID_LIST,                 
00474     OPER_EXPRESSION               
00475   } PdbxStructAssemblyGenItems;
00476 
00478   typedef enum {
00479     PSOL_ID,    
00480     PSOL_TYPE,  
00481     VECTOR_1,   
00482     VECTOR_2,   
00483     VECTOR_3,   
00484     MATRIX_1_1, 
00485     MATRIX_1_2, 
00486     MATRIX_1_3, 
00487     MATRIX_2_1, 
00488     MATRIX_2_2, 
00489     MATRIX_2_3, 
00490     MATRIX_3_1, 
00491     MATRIX_3_2, 
00492     MATRIX_3_3  
00493   } PdbxStructOperListItems;
00494 
00496   typedef enum {
00497      STRUCT_ENTRY_ID,     
00498      PDBX_CASP_FLAG,      
00499      PDBX_DESCRIPTOR,     
00500      PDBX_FORMULA_WEIGHT, 
00501      PDBX_FORMULA_WEIGHT_METHOD, 
00502      PDBX_MODEL_DETAILS,  
00503      PDBX_MODEL_TYPE_DETAILS, 
00504      STRUCT_TITLE         
00505   } StructItems;
00506 
00508   typedef enum {
00509     SC_BEG_AUTH_ASYM_ID,  
00510     SC_BEG_LABEL_ASYM_ID, 
00511     SC_BEG_LABEL_COMP_ID, 
00512     SC_BEG_LABEL_SEQ_ID,  
00513     SC_CONF_TYPE_ID,      
00514     SC_END_AUTH_ASYM_ID,  
00515     SC_END_LABEL_ASYM_ID, 
00516     SC_END_LABEL_COMP_ID, 
00517     SC_END_LABEL_SEQ_ID,  
00518     SC_ID,                
00519   } StructConfItems;
00520 
00522   typedef enum {
00523     SSR_BEG_LABEL_ASYM_ID,     
00524     SSR_BEG_LABEL_COMP_ID,     
00525     SSR_BEG_LABEL_SEQ_ID,    
00526     SSR_END_LABEL_ASYM_ID,     
00527     SSR_END_LABEL_COMP_ID,     
00528     SSR_END_LABEL_SEQ_ID,      
00529     SSR_SHEET_ID,              
00530     SSR_ID,                    
00531     SSR_BEG_AUTH_ASYM_ID,      
00532     SSR_END_AUTH_ASYM_ID,      
00533     } StructSheetRangeItems;
00534 
00536   typedef enum {
00537     DATE,           
00538     PDPOS_ID,       
00539     PDB_ID,         
00540     REPLACE_PDB_ID, 
00541   } PdbxDatabasePDBObsSpr;
00542 
00544   typedef enum {
00545     DPI_NUM,           
00546     DPI_DATE,          
00547     DPI_DATE_ORIGINAL, 
00548     DPI_STATUS,        
00549   } DatabasePDBRevItems;
00550 
00552   typedef enum {
00553     PARH_ORDINAL,         
00554     PARH_REVISION_DATE,   
00555   } PdbxAuditRevisionHistoryItems;
00556 
00558   typedef enum {
00559     PARD_REVISION_ORDINAL, 
00560     PARD_TYPE,             
00561   } PdbxAuditRevisionDetailsItems;
00562 
00564   typedef enum {
00565     PDS_RECVD_INITIAL_DEPOSITION_DATE, 
00566   } PdbxDatabaseStatusItems;
00567 
00569   typedef enum {
00570     ATOM_SITE,
00571     ENTITY,
00572     ENTITY_POLY,
00573     CITATION,
00574     CITATION_AUTHOR,
00575     EXPTL,
00576     REFINE,
00577     PDBX_STRUCT_ASSEMBLY,
00578     PDBX_STRUCT_ASSEMBLY_GEN,
00579     PDBX_STRUCT_OPER_LIST,
00580     STRUCT,
00581     STRUCT_CONF,
00582     STRUCT_SHEET_RANGE,
00583     PDBX_DATABASE_PDB_OBS_SPR,
00584     STRUCT_REF,
00585     STRUCT_REF_SEQ,
00586     STRUCT_REF_SEQ_DIF,
00587     DATABASE_PDB_REV,
00588     PDBX_AUDIT_REVISION_HISTORY,
00589     PDBX_AUDIT_REVISION_DETAILS,
00590     PDBX_DATABASE_STATUS,
00591     DONT_KNOW
00592   } MMCifCategory;
00593 
00595   typedef struct {
00596     mol::ChainType type; 
00597     String details;      
00598     String seqres;       
00599   } MMCifEntityDesc;
00600   typedef std::map<String, MMCifEntityDesc> MMCifEntityDescMap;
00601 
00603   typedef struct {
00604     String biounit_id;                              
00605     std::vector<String> chains;                     
00606 
00607     std::vector<std::vector<String> > operations;   
00608 
00609   } MMCifBioUAssembly;
00610   typedef std::vector<MMCifBioUAssembly> MMCifBioUAssemblyVector;
00611 
00612   typedef std::map<String, std::pair<std::vector<int>, std::vector<String> > >
00613     MMCifCitationAuthorMap;
00614 
00616   typedef struct {
00617     mol::ResNum start;
00618     mol::ResNum end;
00619     String chain_name;
00620   } MMCifHSEntry;
00621   typedef std::vector<MMCifHSEntry> MMCifHSVector;
00622 
00624   typedef struct {
00625     String details;
00626     String method_details;
00627   } MMCifPSAEntry;
00628   typedef std::map<String, MMCifPSAEntry> MMCifPSAMap;
00629 
00630   // members
00631   MMCifCategory category_;
00632   int category_counts_[DONT_KNOW+1]; 
00633   int indices_[MAX_ITEMS_IN_ROW]; 
00634   const IOProfile& profile_;
00635   mol::EntityHandle& ent_handle_;
00636   String restrict_chains_;
00637   bool auth_chain_id_;       
00638   bool seqres_can_;          
00639   mol::ChainHandle curr_chain_;
00640   mol::ResidueHandle curr_residue_;
00641   int chain_count_;
00642   int residue_count_;
00643   int atom_count_;
00644   bool warned_name_mismatch_;
00645   bool warned_rule_based_;
00646   String subst_res_id_; 
00647   bool has_model_;      
00648   int curr_model_;      
00649   std::vector<std::pair<mol::ChainHandle, String> > chain_id_pairs_;
00651   MMCifEntityDescMap entity_desc_map_; 
00652   seq::SequenceList seqres_;
00653   bool read_seqres_;
00654   MMCifInfo info_;      
00655   MMCifCitationAuthorMap authors_map_;
00656   MMCifBioUAssemblyVector bu_assemblies_;
00657   MMCifPSAMap bu_origin_map_; 
00658   MMCifHSVector helix_list_; 
00659   MMCifHSVector strand_list_; 
00660   MMCifInfoStructRefs struct_refs_;
00661   // for storing revisions 
00662   std::map<int, String> revision_dates_;
00663   std::map<int, String> revision_types_;
00664   bool database_PDB_rev_added_;
00665 };
00666 
00667 }}
00668 
00669 #endif
stage/include/ost/io/mol/mmcif_reader.hh
