ó Ú™xTc @ s³dZddlmZddlmZddlmZmZddlmZddlm Z ddl Zddl Z ddl Z ej jdƒZd d eed eddd „Zeeƒd „Zeed„Zdefd„ƒYZdeedd deƒd„Zdefd„ƒYZeeedd eeeƒd„Zd„Zddeƒd„Zdefd„ƒYZdd„Zee jd„Z deeddd„Z!dS(sZ Tools for assembling an ensemble of related structures for local rebuilding by homology. iÿÿÿÿ(tdivision(t+slots_getstate_setstate_default_initializer(tAutotadopt_init_args(tnull_out(teasy_mpNsÔ min_identity = 0.95 .type = float xray_only = True .type = bool single_chain = False .type = bool min_resolution = 3.0 .type = float sort_by_resolution = True .type = bool ligands = None .type = str gffffffî?g@gH¯¼šò×z>c C s|dkrtƒ}nddlm} d} |dk rñt|ƒdkrñx¢|D]—} | jddƒjdƒ} xs| D]k} | jd| d|d|d |ƒ}| dkrÀtgƒ} n| j g|D]}|j ƒ^q̓q{WqSWn| j d |d |d |d|d |d|d |d|ƒ}g}xE|D]=}|j ƒ}| dk rj|| krjq:n|j |ƒq:W|S(sv Finds closely related structures in the RCSB database, optionally limited to structures with specific ligands. iÿÿÿÿ(trcsb_web_servicesit t,tresnametd_maxt xray_onlyt data_onlytsequencet min_identitytexpecttsort_by_resolutiontlogN( tNoneRt mmtbx.wwpdbRtlentreplacetsplittchemical_id_searchtsettupdatetlowertsequence_searchtappend(R Rtmin_resolutionR R RRtligandsRRt allowed_settlig_strtresnamesR tidstpdb_idt matching_idstfiltered((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytfetch_similar_pdb_ids!s@      1    c C sêddl}|jjj|ƒg}x¾|D]¶}y4|jjjd|dtdtjƒƒ\}}Wn=tk r¥}|s„‚qâ|d|IJ|t |ƒIJq,Xt |j ƒƒdkrÏ|d|IJq,|j ||fƒq,W|S( s_ Given a list of PDB IDs, load hierarchies for all of them (ignoring any unknown atoms). iÿÿÿÿNtidtallow_unknownst local_cachesError - skipping %s:is%Warning: %s is multi-MODEL - skipping( tiotbx.pdb.fetchtpdbtfetchtvalidate_pdb_idstload_pdb_structuretTruetostgetcwdt ExceptiontstrRtmodelsR( tpdb_idstfault_tolerantRtiotbxtids_and_hierarchiesR#t pdb_hierarchytxray_structurete((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytload_pdb_modelsNs$  c C sddlm}m}tjj|ƒs.t‚g}xÖtj|ƒD]Å}tjj||ƒ}tjj|ƒrœ|rœ|j t d|d|dt ƒƒqDtjj |ƒrD|rÌ||ƒdkrÌqDn||dt ƒ}|j dkr |j||jfƒq qDqDW|S(sq Load all models in the specified directory, returning a list of file names and iotbx.file_reader objects. iÿÿÿÿ(tany_filetguess_file_typetdir_nametlimit_extensionst recursiveR+t"raise_sorry_if_not_expected_format(tiotbx.file_readerR=R>R0tpathtisdirtAssertionErrortlistdirtjointextendtload_all_models_in_directoryR/tisfilet file_typeRt file_object( R?R@RAR=R>tfile_names_and_objectst file_namet full_patht input_file((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyRJis$     t related_chaincB s eZdZddddgZRS(s} Container for a protein chain (as hierarchy object) and metadata, extracted based on similarity to a target sequence. t source_infotchain_idR9tidentity(t__name__t __module__t__doc__t __slots__(((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyRRƒsg{®Gáz”?c  C sGddl} ddl} g} x"|jƒjƒD]} | jƒr1| jdtdddtƒ} | jj d| d|ƒj ƒ}|j d dgƒ}|||kr?| j j jƒ}| j j jƒ}|j|ƒ| j}|dk rö|}n| j j jd |ƒ}|j|ƒx®| jƒD] }|jƒ}| j j jd |jd |jƒ}xZt|ƒD]L\}}|r¤|d ksk|jjƒdkr¤qkn|j|jƒƒqkW|j|ƒq(W|j ƒ}|j!ƒ|j"ƒj#|j$ƒ|j%ƒƒ| j&t'd|d| jd|d|ƒƒ|r<Pq<q?q1q1W| S(sž Find all chains with at least the specified fractional sequence identity to the target, and extract as related_chain objects with new pdb hierarchies. iÿÿÿÿNtpadtsubstitute_unknowntXt pad_at_starttseq_atseq_bt skip_charsR'tresseqticodeittARSRTR9RU(RcRd((tmmtbx.alignmenttiotbx.pdb.hierarchyt only_modeltchainst is_proteintas_padded_sequenceR/tFalset alignmenttaligntextract_alignmenttcalculate_sequence_identityR+t hierarchytroottmodelt append_modelR'Rtchaint append_chaintresidue_groupst atom_groupst residue_groupRaRbt enumeratetaltloctstriptappend_atom_groupt detached_copytappend_residue_grouptextract_xray_structuretconvert_to_isotropictatomstset_adps_from_scattererst scattererst unit_cellRRR(R9R RStfirst_chain_onlytremove_alt_confstreset_chain_idRtmin_identity_epsilonRtmmtbxR7tresultsRtt chain_seqRlRURqRrRTt new_chainRxRwtnew_rgtkt atom_grouptxrs((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytfind_similar_chainsŠsZ             !     textract_related_modelscB s;eZdZeddeeƒd„Zd„Zd„Z RS(s’ Multiprocessing wrapper for pulling matching chains out of a collection of PDB hierarchies (either from local files or from a PDB mirror). gffffffî?c C s˜t|tƒƒtjd|jdtt|jƒƒd|jƒ|_ xKt |j ƒD]:\}} t| ƒdkrV|d|j|dIJqVqVWdS(Nt fixed_functiterablet processesis no matches for %s( RtlocalsRtpool_mapt examine_modeltrangeRtsources_and_modelstnproct_resultsRy( tselfRšR R…R‡RR›RRŽRŠ((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyt__init__És  cC s+g}x|jD]}|j|ƒqW|S(N(RœRI(RRŠt result_list((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyRŠÚscC sV|j|\}}td|d|jd|d|jd|jd|jddƒ}|S(NR9R RSRR…R‡R(RšR‘R RR…R‡R(Rti_modelRSR9trelated((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyR˜ßs     N( RVRWRXRkRRRRžRŠR˜(((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyR’Äs c C s€td|d|d|ƒ} g| D]\} } | | jf^q"} td| d|d|d|d|d |d | ƒjƒ}|S( sACombines load_all_models_in_directory and extract_related_models.R?R@RARšR RR…R‡R›R(RJRpR’RŠ(R?R R@RAR…R‡Rtreturn_original_filesR›RRNRORMRšR¡((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyt load_related_models_in_directoryës  %cK s7td|d|jdtƒƒƒ}t||jƒS(sÉ Combines load_pdb_models and extract_related_models, with multiprocessing support. Functionally equivalent to load_related_models_in_directory but with models from the PDB (local or remote). R5R(R<tgetRR’RŠ(R5tkwdsRš((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytload_related_models_from_pdbs c C s[ddlm}t|ƒjdkr7|j|ƒ}nt|ƒjdkr^|j|ƒ}ng}xðttƒD]â\}}t} d} |dk rÚ|dkrªt } q%t |jƒƒ} | t |ƒkr%t } q%nK|dk r%|dk r%t |jƒƒ} | t |ƒkr"t } q"q%n| rF|d||| fIJqq|j |ƒqqW|S(s• Return a list of indices for those crystal symmetries which either match the only_space_group parameter, or do not match exclude_space_group. iÿÿÿÿ(tsgtbxtinfos# %s is in space group %s, skippingN( tcctbxR§ttypeRVtspace_group_infoRyt symmetriesRkRR/R3R( tcrystal_symmetriest source_infostexclude_space_grouptonly_space_groupRR§t selectionRŽtsymmtskiptsg((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytselection_by_symmetrys0       t ensemblercB sJeZdZdedeedeƒd„Zd„Zd„Z d„Z RS(s½ Superpose a collection of single-chain models on a reference chain. This is intentionally very limited in function, but flexible with respect to the choice of atoms to superpose. gà?c C s$t|tƒƒ|jd|ƒdS(Ntatom_selection_string(RR–trealign( Rtreference_hierarchytrelated_chainsR·t backbone_onlyt calpha_onlyR›t sieve_fitt frac_discardR((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyRž<s cC sº||_|jjƒ}|jƒ|jdk rS|j rJ|jtks†t‚n3|jrhd|_n|jr}d|_n d|_ddl m }|jj ƒ}|j |jƒ|_ |j jtƒdksØt‚g|_|jƒ}|jjƒjƒ}xŒ|jƒD]~}|jƒjƒ}xc|D][} |j | jsJq.n|jƒ} |jjd| | jjƒfƒ|j| jƒq.WqWt|ƒdks©t‚|jƒ} | j|ƒ|_tjdt t|j!ƒƒd |j"d |j#ƒ} g|_$x°t%| ƒD]¢\} }|j!| j&}|dkri|j'd |j!| j(|j!| j)fIJqn|j$j| ƒ|jƒ}|j*j+|jƒ|j,j+}|j-|ƒqWdS( Nsname CAs0name CA or name CB or name C or name N or name Otalliÿÿÿÿ(tflexis%s %sR”R“R•sNo LSQ fit for model %s:%s(.R·R¹Rt reset_i_seqRR¼R»R/RFtscitbx.array_familyRÀtatom_selection_cacheR±tatom_selectiontcountt atoms_reftsize_tRgt only_chainRvtonly_atom_groupti_seqtresidRtnameR{Rt extract_xyztselecttreference_sitesRR—R™Rºt align_modelR›tselection_movedRyR9RRSRTtrtelemstttset_xyz(RR·t ref_atomsRÀt sel_cachetref_selt ref_chainRxtrg_atomstatomRËt sites_reft sites_movedRŽtlsq_fitRpt pdb_atomst sites_cart((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyR¸IsV  "         #       !  cC sÿddlm}ddlm}|j|j}|jƒ}|jƒ|jƒ}|j |j ƒ}|j ƒj ƒ}|j ƒ} |j ƒ} x¤|jƒD]–} x| jƒjƒD]y} || jsÍq´n| jƒ} d| | jjƒf}||jkr´| j| jƒ| j|jj|ƒƒq´q´Wq›Wt| ƒdkr`|j dk s\t‚dSt| ƒdkrŠt| ƒt| ƒkst‚|jƒ}|j| ƒ}|jj| ƒ}|jrå|jd|d|d|jƒS|jd |d |ƒSdS( Niÿÿÿÿ(RÀ(t superposes%s %sit sites_fixedt sites_movingR¾RÏt other_sites( RÂRÀt scitbx.mathRáRºR9RRÁRÃR±R·RgRÈRÇRvRÉRÊRËRÌR{RÆRtindexRRRFRÍRÎRÏR½R¾tleast_squares_fit(RR RÀRáthierarchy_movingt mov_atomsR×tmov_atom_selectiont mov_chaintsel_reftsel_movRxRÛRËtref_nametxyz_movt sites_movRÜ((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyRÐzsD       $0     cC s;g}x(|jD]}|j|j|jƒqWt|ƒS(N(RÑRRºR9tjoin_hierarchies(Rt hierarchiesRŽ((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytas_multi_model_hierarchyŸsN( RVRWRXRRRkRRžR¸RÐRó(((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyR¶6s 1 %c C s²ddl}|jjjƒ}xt|ƒD]\}}|d}|dk rZ||}n|jjjdt|ƒƒ}|j|ƒ|j ƒj ƒj ƒ}|j |ƒq+W|S(NiÿÿÿÿiR'( RfR+RpRqRyRRrR3RsRgRÈR}Ru( Ròt model_idsR7RqRŽRptmodel_idRrRt((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyRñ¥s     c C s¦|dkrotd|d|jd|jd|jd|jd|ƒ}|jdk r¯g|jD]}|jƒjdƒ^q^}x-|D]"} | |kr†|j | ƒq†q†Wnt |ƒdkrÅdS|d t |ƒIJd} xB| t |ƒkr#|d d j || | d !ƒIJ| d 7} qâWt d |d|d|j d|jd|jƒjƒjd|d|ƒ} n tƒ‚| dk sŠt‚td|d| d|jƒS(NR RRRR RRis%d PDB IDs retrieved:s %sRiR5R…R‡R›R¹Rº(RR&RRRR t exclude_idsRRtremoveRRHR¦t single_chainRgRÈR'tNotImplementedErrorRFR¶R›( R¹tsearch_directoryR tparamsR›toutR5tsRötother_idtiRº((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytextract_and_superpose²sD      +  "    RdRc sàddlm}ddl‰|j‰g}‡‡fd†} ‡‡fd†} xŠ|jƒjƒD]v} | jƒrbd} | jƒ} xO| t| ƒkrÔ| | | dƒ}|dk rÇ|g}x~t dtˆƒƒD]g}| |}|t| ƒkrg}Pn| | d|d|ƒ}|dk r8|j |ƒqØg}PqØWt|ƒdkrÇt|ƒtˆƒksst ‚| ƒ}|jƒj ƒ}x¨t|ƒD]š\}}|jƒj}|jƒj}|dk râd ||}n|dk r÷|}nˆjjjd |d |ƒ}|j|ƒ|j|jƒƒq›Wd}|jƒ}|rc|jd ƒ}n|r{|jd ƒ}n|dk r´|jtƒdks¢t ‚|j|ƒ}n|j |ƒqÇn| d7} q‰WqbqbW|S(Niÿÿÿÿ(tamino_acid_codesc sZˆjjjƒ}ˆjjjƒ}ˆjjjdˆƒ}|j|ƒ|j|ƒ|S(NR'(R+RpRqRrRtRsRu(RqRrRt(R7R‡(sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytmake_hierarchyês   c sUxN||jƒD]<}ˆj|jƒ}|ˆ|ksIˆ|dkr|SqWdS(Nt.R\(RR\(RwR¤R R(Rvti_resRÊtnext_atom_grouptseq_code(t aa_3_as_1R (sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytfind_next_matching_atom_groupñs  iiRRÊs%4dRaRbs0name C or name O or name N or name CA or name CBsnot (element H or element D)(t iotbx.pdbRRftone_letter_given_three_letterRgRhRiRvRRR™RRFRÈRytparentRaRbR+RpRxR~R|R}RÃR±RÅR/RÎ(R9R t renumber_fromtmainchain_onlytremove_hydrogensR‡t reset_icodeRt fragmentsRRRtRRvRtfragmentRÊtj_resRRqRŒti_groupRaRbtnew_residue_grouptselR×((RR7R‡R sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pyt%extract_peptide_fragments_by_sequenceÞsn                  ("RXt __future__RtlibtbxRRRt libtbx.utilsRRt libtbx.philR0tsystphiltparset master_philR/RkRR&R<RJRRR‘tobjectR’R£R¦RµR¶RñtstdoutRR(((sI/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/building/make_library.pytsf     $  2* !o )