B bR@spdZddlZddlZddlmZddlmZddlmZddl m Z ddl m Z Gdd d Z Gd d d Z dS) zmmCIF parsers.N) as_handle) MMCIF2Dict)StructureBuilder)PDBConstructionException)PDBConstructionWarningc@sBeZdZdZdddZddZdd Zd d Zd d ZddZ dS) MMCIFParserz1Parse a mmCIF file and return a Structure object.NFcCs8|dk r||_nt|_d|_d|_d|_t||_dS)aCreate a PDBParser object. The mmCIF parser calls a number of standard methods in an aggregated StructureBuilder object. Normally this object is instanciated by the MMCIParser object itself, but if the user provides his/her own StructureBuilder object, the latter is used instead. Arguments: - structure_builder - an optional user implemented StructureBuilder class. - QUIET - Evaluated as a Boolean. If true, warnings issued in constructing the SMCRA data will be suppressed. If false (DEFAULT), they will be shown. These warnings might be indicative of problems in the mmCIF file! Nr)_structure_builderrheader line_counterbuild_structureboolQUIET)selfstructure_builderr r2lib/python3.7/site-packages/Bio/PDB/MMCIFParser.py__init__szMMCIFParser.__init__c CsVt>|jrtjdtdt||_|||j | WdQRX|j S)zReturn the structure. Arguments: - structure_id - string, the id that will be used for the structure - filename - name of mmCIF file, OR an open text mode file handle ignore)categoryN) warningscatch_warningsr filterwarningsrr _mmcif_dict_build_structurerZ set_header _get_header get_structure)r structure_idfilenamerrrr2s   zMMCIFParser.get_structurecCs$||kr ||d}d|kr |S|S)Nr?r)rkeydictZdefltZrsltrrr _mmcif_getEs  zMMCIFParser._mmcif_getc Cs^|j}xR|D]J}||}y |d}Wnttfk r@w YnX|dkr ||j|<Pq WdS)Nrr)rget TypeError IndexErrorr )rZ target_keykeysZmdrvalitemrrr_update_header_entryLs    z MMCIFParser._update_header_entrycCsddddddd|_|ddddg|ddg|d d d g|d d g|ddg|ddddg|jddk ryt|jd|jd<Wntk rd|jd<YnX|jS)N)nameheadidcodedeposition_datestructure_method resolutionr,Z _entry_idz_exptl.entry_idz_struct.entry_idr*z _struct.titler+z_struct_keywords.pdbx_keywordsz_struct_keywords.textr-z3_pdbx_database_status.recvd_initial_deposition_dater.z _exptl.methodr/z_refine.ls_d_res_highz_refine_hist.d_res_highz _em_3d_reconstruction.resolution)r r(float ValueError)rrrrrXs2   zMMCIFParser._get_headerc> Csddh}|j}|d}|d}|d}y |d}Wntk rJd}YnX|d}dd |d D} d d |d D} d d |dD} |d} |d} |d}|d}|d}ydd |dD}Wn4tk rd}Yntk rtddYnXy8|d}|d}|d}|d}|d}|d}d}Wntk rPd}YnXd|krf|d}n|d }d}d}d}|j}|||d!d"}d"}xtdt|D]} | | yt || }!Wn*tk r|| }!t d#t YnX| | }"| | }#| | }$|| }%|| }&| | }'|'|kr8d!}'t || }(| | })|)|krZd!})|| }*yt|| }+Wn tk rtd$dYnXyt|| },Wn tk rtd%dYnX|| }-|-d&kr|%d'ks|%d(krd)}.nd*}.nd!}.|.|(|)f}/|dk rH|| }0||0krR|0}|d7}|||d}d}d}n ||||&krr|&}||d}d}||/ks||%kr|/}|%}||%|.|(|)t|"|#|$fd+}1|r|| nd}2|j|*|1|+|,|'|*|!|2d,|dkr| t|kr|| || || || || || f}3d-d |3D}4t|4d+}5||5qWyt|d.d}6t|d/d}7t|d0d}8t|d1d}9t|d2d}:t|d3d};t|6|7|8|9|:|;fd+}<|d4d}=|=dd"}=|=dkrt||=|<Wntk rYnXdS)5N.rz _atom_site.idz_atom_site.label_atom_idz_atom_site.label_comp_idz_atom_site.type_symbolz_atom_site.auth_asym_idcSsg|] }t|qSr)r0).0xrrr sz0MMCIFParser._build_structure..z_atom_site.Cartn_xcSsg|] }t|qSr)r0)r3r4rrrr5sz_atom_site.Cartn_ycSsg|] }t|qSr)r0)r3r4rrrr5sz_atom_site.Cartn_zz_atom_site.label_alt_idz_atom_site.pdbx_PDB_ins_codez_atom_site.B_iso_or_equivz_atom_site.occupancyz_atom_site.group_PDBcSsg|] }t|qSr)int)r3nrrrr5sz_atom_site.pdbx_PDB_model_numzInvalid model numberz_atom_site_anisotrop.U[1][1]z_atom_site_anisotrop.U[1][2]z_atom_site_anisotrop.U[1][3]z_atom_site_anisotrop.U[2][2]z_atom_site_anisotrop.U[2][3]z_atom_site_anisotrop.U[3][3]rz_atom_site.auth_seq_idz_atom_site.label_seq_id zBPDBConstructionWarning: Some atom serial numbers are not numericalzInvalid or missing B factorzInvalid or missing occupancyHETATMZHOHZWATWHf) serial_numberelementcSsg|] }t|qSr)r0)r3_rrrr5sz_cell.length_az_cell.length_bz_cell.length_cz_cell.angle_alphaz_cell.angle_betaz_cell.angle_gammaz_symmetry.space_group_name_H-M)rKeyErrorr1rrinit_structureinit_segrangelenset_line_counterr6rwarnrr0 init_model init_chain init_residuenumpyarrayupper init_atom set_anisou ExceptionZ set_symmetry)>rr _unassigned mmcif_dictatom_serial_list atom_id_listresidue_id_list element_list chain_id_listx_listy_listz_listalt_list icode_list b_factor_listoccupancy_listfieldname_list serial_list aniso_u11 aniso_u12 aniso_u13 aniso_u22 aniso_u23 aniso_u33 aniso_flag seq_id_listcurrent_chain_idcurrent_residue_idcurrent_resnamercurrent_model_idcurrent_serial_idiserialr4yzresnamechainidaltloc int_resseqicoder* tempfactor occupancy fieldname hetatm_flagresseq serial_idcoordr@u mapped_anisou anisou_arrayabcZalphaZbetaZgammaZcellZ spacegrouprrrr}s                         zMMCIFParser._build_structure)NF) __name__ __module__ __qualname____doc__rrr!r(rrrrrrrs  %rc@s*eZdZdZd ddZddZdd ZdS) FastMMCIFParserz2Parse an MMCIF file and return a Structure object.NFcCs2|dk r||_nt|_d|_d|_t||_dS)arCreate a FastMMCIFParser object. The mmCIF parser calls a number of standard methods in an aggregated StructureBuilder object. Normally this object is instanciated by the parser object itself, but if the user provides his/her own StructureBuilder object, the latter is used instead. The main difference between this class and the regular MMCIFParser is that only 'ATOM' and 'HETATM' lines are parsed here. Use if you are interested only in coordinate information. Arguments: - structure_builder - an optional user implemented StructureBuilder class. - QUIET - Evaluated as a Boolean. If true, warnings issued in constructing the SMCRA data will be suppressed. If false (DEFAULT), they will be shown. These warnings might be indicative of problems in the mmCIF file! Nr)rrr r r r )rrr rrrr4s zFastMMCIFParser.__init__c CsRt:|jrtjdtdt|}|||WdQRXWdQRX|jS)zReturn the structure. Arguments: - structure_id - string, the id that will be used for the structure - filename - name of the mmCIF file OR an open filehandle r)rN) rrr rrrrrr)rrrZhandlerrrrRs    zFastMMCIFParser.get_structurec@ Csddh}d\}}gg}}gg}} x|D]} | drLd}|| q*| drjd}|| q*|r~| dr~d}q*|r| drd}q*|r|| q*|r*| | q*Wtttj|} tttj| } tt|| } | tt|| | d }| d }| d }y | d }Wnt k r<d}YnX| d }dd| dD}dd| dD}dd| dD}| d}| d}| d}| d}| d}ydd| dD}Wn8t k rd}Yn t k rt ddYnXy8| d}| d}| d}| d }| d!} | d"}!d#}"Wnt k rFd$}"YnXd%| kr\| d%}#n| d&}#d}$d}%d}&|j }'|' ||'d'd(}(d(})x^td$t|D]J}*|'|*||*}+||*},||*}-||*}.||*}/||*}0||*}1|1|krd'}1t|#|*}2||*}3|3|krd'}3||*d)}4yt||*}5Wn t k rZt d*dYnXyt||*}6Wn t k rt d+dYnX||*}7|7d,krd-}8nd'}8|8|2|3f}9|dk r||*}:|)|:kr|:})|(d#7}(|'|(|)d}$d}%d}&n |'|(|$|0kr |0}$|'|$d}%d}&|%|9ks4|&|/krL|9}%|/}&|'|/|8|2|3t|,|-|.fd.};|rl||*nd}<|'j|4|;|5|6|1|4|+|t|>d.}?|'|?qWdS)1Nr2r)FFz _atom_site.Tz_atom_site_anisotrop.#Fz _atom_site.idz_atom_site.label_atom_idz_atom_site.label_comp_idz_atom_site.type_symbolz_atom_site.auth_asym_idcSsg|] }t|qSr)r0)r3r4rrrr5sz4FastMMCIFParser._build_structure..z_atom_site.Cartn_xcSsg|] }t|qSr)r0)r3r4rrrr5sz_atom_site.Cartn_ycSsg|] }t|qSr)r0)r3r4rrrr5sz_atom_site.Cartn_zz_atom_site.label_alt_idz_atom_site.pdbx_PDB_ins_codez_atom_site.B_iso_or_equivz_atom_site.occupancyz_atom_site.group_PDBcSsg|] }t|qSr)r6)r3r7rrrr5sz_atom_site.pdbx_PDB_model_numzInvalid model numberz_atom_site_anisotrop.U[1][1]z_atom_site_anisotrop.U[1][2]z_atom_site_anisotrop.U[1][3]z_atom_site_anisotrop.U[2][2]z_atom_site_anisotrop.U[2][3]z_atom_site_anisotrop.U[3][3]r8rz_atom_site.auth_seq_idz_atom_site.label_seq_idr9r:"zInvalid or missing B factorzInvalid or missing occupancyr;r=r>)r?r@cSsg|] }t|qSr)r0)r3rArrrr5s) startswithappendstripzipmapstrsplitr updaterBr1rrrCrDrErFrGr6r0rIrJrKrLrMrOrP)@rrZ filehandlerRZ read_atomZ read_aniso_fieldsZ_recordsZ_anisofZ_anisorslineZ _record_tblZ _anisob_tblrSrTrUrVrWrXrYrZr[r\r]r^r_r`rarbrcrdrerfrgrhrirjrkrlrrmrnrorpr4rqrrrsrtrurvrwr*rxryrzr{r|r}r~r@rrrrrrrds                          z FastMMCIFParser._build_structure)NF)rrrrrrrrrrrr1s r)rrLrZBio.FilerZBio.PDB.MMCIF2DictrZBio.PDB.StructureBuilderrZBio.PDB.PDBExceptionsrrrrrrrrs