əxTc@ sjdZddlmZddlmZddlmZddlmZddlmZddl m Z ddlZ dd l m Z dd lmZmZdd lmZmZdd lmZdd lmZddlmZmZmZddlmZmZmZddl m!Z!m"Z"ddlm#Z#ddl$Zddl%Zddl&m'Z'ddl(Z(ddl)Z)ddl*Z*dej+Z,dZ-e.d\Z/Z0Z1Z2dddgZ3de4fdYZ5de4fdYZ6de4fdYZ7de j8fd YZ9d!e4fd"YZ:d#e4fd$YZ;d%Z<d&ee=e>e>d'Z?d(Z@dS()s{ Examines a structure for metal ions. Can iterate over all atoms, examining their density and chemical environment to determine if they are correctly identified, or if there are better candidate ions that they can be replaced with. See mmtbx.ions.build to actually modify the structure, the code in this module handles ion identification, but only prints out messages to a log. i(tdivision(tfind_coordination_geometry(t environment(thalides(tutils(t validationN(tcommon_residue_names_water(tsasakithenke(tcrystaltadptbx(tflex(tcol(tmake_sub_headert format_valuet framed_output(t group_argstadopt_init_argstAuto(tnull_outtSorry(teasy_mp(tsqrtsA require_valence = False .type = bool .help = Toggles the use of valence calculations .short_caption = Require good bond valence ambiguous_valence_cutoff = 0.5 .type = float .help = "Cutoff to uniquely select one of many metals by comparing its observed and expected valence" d_min_strict_valence = 1.5 .type = float .short_caption = Resolution limit for strict valence rules anom_map_type = *residual simple llg .type = choice .help = Type of anomalous difference map to use. Default is a residual map, showing only unmodeled scattering. find_anomalous_substructure = Auto .type = bool use_phaser = True .type = bool .help = Toggles the use of Phaser for calculation of f-prime values. .short_caption = Use Phaser to calculate f-double-prime aggressive = False .type = bool .help = Toggles more permissive settings for flagging waters as heavier elements. Not recommended for automated use. map_sampling_radius = 2.0 .type = float .input_size = 80 .short_caption = Sampling radius water .short_caption = Water filtering .style = box auto_align { min_2fofc_level = 1.8 .type = float .input_size = 80 .help = Minimum water 2mFo-DFc map value. Waters below this cutoff will not be analyzed further. .short_caption = Min. allowed 2mFo-DFc map value max_fofc_level = 3.0 .type = float .input_size = 80 .help = Maximum water mFo-DFc map value .short_caption = Max. expected mFo-DFc map value max_anom_level = 3.0 .type = float .input_size = 80 .help = Maximum water anomalous map value .short_caption = Max. expected anomalous map value max_occ = 1.0 .type = float .input_size = 80 .help = Maximum water occupancy .short_caption = Max. expected occupancy min_b_iso = 1.0 .type = float .input_size = 80 .help = Minimum water isotropic B-factor .short_caption = Min. expected B-factor fp_max = 1.0 .type = float fpp_max = 0 .type = float max_stddev_b_iso = 5 .type = float .input_size = 80 .short_caption = Max. standard deviations below mean B-iso min_frac_b_iso = 0.2 .type = float .input_size = 80 .short_caption = Min. fraction of mean B-iso min_frac_calpha_b_iso = 0.75 .type = float max_frac_calpha_2fofc = 1.2 .type = float min_2fofc_coordinating = 0.9 .type = float .help = Minimum 2mFo-DFc for waters involved in coordination shells. } phaser .short_caption = Phaser options .caption = These parameters control the identification of anomalous scatterers using Phaser substructure completion. .style = box auto_align { llgc_ncycles = None .type = int .short_caption = Number of cycles distance_cutoff = 1.5 .type = float .input_size = 80 .short_caption = Min. required separation from other atoms distance_cutoff_same_site = 0.7 .type = float .input_size = 80 .short_caption = Max. separation from mapped atom fpp_ratio_min = 0.2 .type = float .input_size = 80 .help = Minimum ratio of refined/theoretical f-double-prime. .short_caption = Min. f-double-prime ratio fpp_ratio_max = 1.1 .type = float .input_size = 80 .help = Maximum ratio of refined/theoretical f-double-prime. .short_caption = Max. f-double-prime ratio } chloride .short_caption = Chloride ions .style = box auto_align { %s } cC stjjtdtS(Ntprocess_includes(tlibtbxtphiltparsetion_identification_phil_strtTrue(((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytion_master_philsitMGtCAtMNtmanagercB speZdZd!d!d!ded!dZdZdZd!d!edZ dZ dZ dZ e jd Zd Zd d eed ZdZdZdZdZdZdZdZd ddZd!edZdZdZdedZeeee jdZdZ e jeedZ!e jedZ"e jed!d Z#RS("s[ Wrapper object for extracting local environment and adding or modifying scatterers. ic  C s| dkrt} n||_i|_||_||_||_d|_t|_ d|_ d|_ d|_ t |dd|_i|_i|_d|_|dk r|jj|_n|jd|d|d|d| |jt} |dk r|j} n| tkr=|jjo7|dk } n| r|dkrbtddn%|jjstddn|jrtjjd r| d IJtj} td |d|d |d |d| d|jj} tj} | d| | IJ| dkr"| dIJq| j |_t!|jdkrS| dIJq| dt!|jIJ|j"d| d t#qt#|_ |j$d| ndS(Ntuse_svmt pdb_hierarchytxray_structuret connectivitytlogsWavelength required when s!find_anomalous_substructure=True.sAnomalous data required when tphasers+ Running Phaser substructure completion...tfmodelt wavelengthtverbosetn_cycless time: %.1fss/ ERROR: Phaser substructure completion failed!is No anomalous scatterers found!s %d anomalous scatterers found(%tNoneRR(tatoms_to_propstparamsR)tnproctphaser_substructuretFalsetflag_refine_substructuretfpp_from_phaser_ax_sitestsite_fptsite_fdptgetattrR"t _map_valuest_map_gaussian_fitstanomalous_flagtf_obstupdate_structuret update_mapstfind_anomalous_substructureRRt use_phaserRtenvt has_modulettimetfind_anomalous_scatterersR't llgc_ncyclestatomstlentanalyze_substructureRtrefine_anomalous_substructure(tselfR(R#R$R.R)R%R/R*R&trefine_substructurett1t phaser_resulttt2((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt__init__sv                              cC sQ|jj}djdjgtD]}d|^q"}|j|jS(sx Fetchs the selection for all waters in the model. Returns ------- scitbx.array_family.flex.size_t s!({}) and element O and altloc ' 's or sresname (R#tatom_selection_cachetformattjointWATER_RES_NAMESt selectiont iselection(RHt sel_cachetitsel_str((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytwater_selections)c C sddlm}|jj}|jd|d|jd|jdtdtdtd |}|jj }t j |j t |_t j|j d |_t j|j d |_xY|D]Q}xH|jD]=}||}t|j|<|j|j|<|j|j|= 3.0 sigmas+ %d waters have anomalous map >= 3.0 sigmaN(RR tmaxRRRRRRWREt map_statstfofcRR,R9(RHR[tn_water_fofc_peakstn_water_anom_peaksRRhR((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt"show_current_scattering_statisticss:            cC s%|jjj}||jjkS(s Checks whether the resolution of a model is high enough to use more strict thresholds for ion valences. Returns ------- bool (R(R:td_minR.td_min_strict_valence(RHR((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytget_strict_valence_flags g@g?c C sT|t|jkst|jj|d \}}}|d kr|jjd|}|j}|j }|||f|j|scS s|jS(N(R(R((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRs(Rtt fractionalizeR t vec3_doubleR$Rtprocess_sites_fracRR tneighbors_fast_pair_generatorR^R`Rhtj_seqt get_rt_mx_it get_rt_mx_jtinversetmultiplyt orthogonalizeRRtdist_sqRRtsort(RHRRtdistance_cutoff_same_siteRRnRtpair_generatortn_xrayt same_atomst other_atomstpairtrt_mx_itrt_mx_jtrt_mx_jit new_site_fracRtatom_seqt site_info((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt_find_atoms_near_sitesB       !c C sd}|jdk sttj|jjd|_|dksO| r[t}nd}t dd|x|jD]w}|||IJ|j |j d|j j jd|j j j\}}||||}}t|dkrD|d ||IJx|D]-}|d |j|jj|jfIJqWq~t|d kr|d |||j|djjfIJt|j|dj<|j|j|djBsRcS s|jS(N(R(R((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRFsi( t collectionsRRhRtvaluestanyRRtremoveR(t site_infosRti_seqsRtvalRU((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt_filter_site_infos-s    s'Analyzing Phaser anomalous substructureR[RR is No match for %ss %s (distance = %.3f)is %s maps to %ss ambiguous results for %s:s %sRN(R0R,RsR RbRpR`R3RR RRR.R'RR RERhtid_strRRRaR( RHR&R*RR#RR R t other_atomt same_atom((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRFs8     '' &cC sQ|j|sdS|jdkr+|j|S|j|}|dkrMd}n|S(s Retrieve the refined f'' for a site. Because this can come from either the built-in anomalous refinement or Phaser, it is handled differently than the f' retrieval. Parameters ---------- i_seq : int Returns ------- float or None iN(RaR,R3R5(RHRhtfpp((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytget_fppds    cC s,|jdks|j| r!dS|j|S(s Retrieve the refined f' for a site. Parameters ---------- i_seq : int Returns ------- float or None N(R4R,Ra(RHRh((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytget_fpzs tCLc C s0|j|}|jj}|jtjjks:t|jdk rf|j |j|j krft S|j |dddt }tjd|d|d|jd|jjd|jd |jd |jjd |}|j|}|j|} t } | dk r,| jd kr,| jd kr,t} n|S(sS Given a site, analyze the nearby atoms (taking geometry into account) to guess whether it might be a halide ion. Among other things, halides will often be coordinated by amide hydrogens, and not in close contact with negatively charged atoms. Note that this procedure has a very high false positive rate when used by itself, so additional information about the electron count (map, occ, b) is absolutely essential. Parameters ---------- i_seq : int element : str assume_hydrogens_all_missing : bool, optional Returns ------- bool Rg@RRhRRpRnRtR%R.tassume_hydrogens_all_missingg?g?N(RpR$RnRRRtHALIDESRsRR,tbRR1RRtis_favorable_halide_environmentRtR%R.tchlorideRRRtstandard_deviationR( RHRhRR+RRnt nearby_atomstfavorable_environmenttpait map_variancetgood_map_falloff((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytlooks_like_halide_ions6          c' sOj|}|tk}|r.tjj}n9t|tsLt|trg|jddj }ng|D]}|j j ^qn}|dgkrg}n|j j j j } | tkstj|d:} jdjjdjj} | tkrd:Sj s(tgt} j} tj}jd:k rmtjj}nx |D]}tjjj |}|d:kr|tjj!krt"} qtqt#d|n| t$t%gkr|d;krqtn| r|t&krqtnt'j(|j j)|j*}|d krKj+d krKqtnj,t|d }|dkr| t$krqtnj-|qtWt"fd}|dt"\}}t"}t.|dkrjj/ r|dt\}}t.|dkrt}qnt}|p0| t0kp0| sst1|j2tjj!}g}x|D]}j3dtjj4d|dddj5djj6j7djj6j8}|d:k r|dkr|jj6j7ks[|jj6j8krq[qnj9d|d|r[|j-|q[q[Wt.|dk}|g|D]'}tjj4d|dddf^qB7}nt.|dkrjj/ rgD]!}j:t|r|^q}x,|D]$}t'j(|j;j)} d}!j,d:k rj,| |j*}!n|j;tjj<kr| t0kr| rj=|rj>|dd}"j?d}#|"d krd|#kodknr|j-|dfqq|j;d<krU| t0krU| rUj=|rUj>|dd"d#d$g}"j@d%d"}$j?d%d"}#|"|$dkr|#d&kr|j-|dfqq| d'kr| t0kr|!d(kr|!d)ksjAd*kr| rj=|d+jB rj?d}#|#d,kr|j-|dfqqqWt.|d krjB rjCt|}%|djDkrjE|%kr|%j2jFjGg rtH|jItHtH|qqnt.|d kr|dd} n| r| rjJd|d-|d.t"d/|}&|&jKd:k r|&SntLd0d1d2|d3|d4| d5| d6|d7jjMd8|d9| dj5d.| S(=s+ Examines the environment around a single atom to determine if it is actually a misidentified metal ion. no_final is used internally, when candidates is not Auto, to try all of the Auto without selecting any as a final choice when none of the elements in candidates appear to probable ion identities. Parameters ---------- i_seq : int debug : bool, optional candidates : list of str, optional no_file : bool, optional out : file, optional Returns ------- mmtbx.ions.identify.water_result t,t tXR.t aggressivesElement '%s' not supported!tNARtFtNEi g?ig?c  sg}g}xD]}jd|djd|jd|djdjjjdjjjt|}jd|r|j |j |fq|j |j |fqW||fS(s Parameters ---------- require_valence : bool, optional Returns ------- list of tuple of mmtbx.ions.metal_parameters, float list of tuple of mmtbx.ions.metal_parameters, float t ion_paramsR)trequire_valencet fpp_ratio_mint fpp_ratio_maxtidentity( tcheck_ion_environmentR)tcheck_fpp_ratioR.R'R@RAtstrtis_correctly_identifiedRtscore(R?t reasonablet unreasonablet elem_paramsRB(t atom_propstfiltered_candidatesRH(sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyttry_candidates$s"       R?iR>RtchargeiR)R@RAR*RhRg@gffffff@iitKRg333333@t exclude_atomsRRiig?g?i tignore_valenceitdebugtno_finalR[RKRLtmatching_candidatestrejected_candidatestnuc_phosphate_sitet atom_typetlooks_like_halidetambiguous_valence_cutofft valence_usedt final_choiceN(R;RR<R=(ROR(NRpRRRt DEFAULT_IONSt isinstanceREtunicodetreplacetsplitRRRRRQRsR-R,t get_atom_typeR.RR:t WATER_POORR"R1t$looks_like_nucleotide_phosphate_sitetsystmaxintRR Rtservertget_metal_parametersR,RRtWATERt LIGHT_IONtNUC_PHOSPHATE_BINDINGRRt atomic_numberRNR'testimated_weightRRER?t HEAVY_IONtsett intersectionRDtmetal_parametersR)R'R@RAR6thas_compatible_ligandsRtTRANSITION_METALStis_compatible_sitet!number_of_favored_ligand_residuestnumber_of_atoms_within_radiustnumber_of_backbone_oxygensRcRt inaccuraciest fpp_ratiostBAD_FPPt BAD_GEOMETRYt NO_GEOMETRYt _PRINT_DEBUGR$t analyze_waterR[t water_resultRY('RHRhRRt candidatesRSR[Rtauto_candidatesRURR[RWthalide_candidatesRVtmax_carbon_fo_map_valuetsymboltelemtn_elect mass_ratioRMRHRIRZRXtcandidate_halidestfiltered_halidesRt fpp_ratiothalideR.t compatibleR>Rkt weight_ratiot n_good_restn_total_coord_atomst n_bb_oxygenRwtresult((RKRLRHsA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyR}s  %        "!      5 !     (              c C s|jj}g}x|jD]}x|jD]}|j}t|dkr_q5nxg|D]_}|jjj}|t krf|j }t|dkr|j |dj qqfqfWq5Wq"W|S(Nii( R#t only_modeltchainstresidue_groupst atom_groupsRERRRRQRDRRh( RHtmodelt water_i_seqstchaint residue_groupRt atom_groupRRD((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt_extract_waterss    &c  s8j}|dIJ|dt|IJt|dkr@dStjj}g}tfd|D_|dkrL|dIJx|D]}tj}jd|d|d |d |} | dk r9| j d |d|j |} | j dk r9| j r9|j|| j g| jfq9ntj} qWn|d |IJ|dIJtd d|d |d |} tjd | d|d|} x| D]z}|dkrqn|\}}}|dk r||IJn|dk rj |} |j||g| jfqqWt|dS(s Iterates through all of the waters in a model, examining the maps and local environment to check their identity and suggest likely ions where appropriate. Parameters ---------- out : file, otpional debug : bool, optional candidates : list of str, optional Returns ------- list of tuple of int, mmtbx.ions.metal_parameters i_seq and metal parameters which indicate that waters of i_seq would be better changed to that new metal identity. Rs %d waters to analyzeiNc3 s$|]}|t|fVqdS(N(tAtomProperties(RRh(RH(sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pys siRhRRRR[s# Parallelizing across %d processesR!t fixed_functargst processescS st|d|dS(Ni(tcmp(taR-((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRs(RWRERt get_processesR/tdictR-RAR}R,t show_summaryRR[RSRRt_analyze_water_wrappertpool_maptsorted(RHR[RRRtwatersR/Rt water_i_seqRJt water_propsRRLR}tresultsRR[t result_str((RHsA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytanalyze_waterss\                    #c C sC|j|}tjjj|j}tjjj|}|d k r|jd|j j }|j d|d|j d|j j |jd|d|j d|j jjd|j jjn|tjjkr |j}t|j|<|jd|d|s?|j|j|jq?ntd ||jjf|S( s Examines one site in the model and determines if it was correctly modelled, returning a boolean indicating correctness. Parameters ---------- i_seq : int out : file, optional debug : bool, optional R.R>R)R?R@RARhRsElement '%s' not supported: %sN(R-RRRft get_elementRRgR,RaR.RRCR)R?RDR'R@RAR,RBRnRwR6taddt BAD_HALIDERtformat_atom_record( RHRhR[RRRKRRJRWRB((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt validate_ions,        c s|dkrg}xjjD]}x|jD]}x|jD]}xw|jD]i}|jjtj j kr[|j } t | dkst x!| D]} || jf7}qWq[q[WqHWq5Wq"Wn-jj} | jdj|j}t |dkrdSg} tfd|D_x?|D]7} jd| d|}| j||jfqFWjj}d#}d}t|dddd}|||IJ|dddIJx| D]\}jj} || }d}}|jjr(|j}|j}nt j!|j"dj#}j$}d}||jj%dt&dd!|djj'|d||dj(|dj)|d||d||dj*dkrdn4|t+j,tj j-j.jj/j*j|dj0j1||dj2j1|f IJ|s|d j3fd!j4|DIJn|d"IJqW|j5dS($s Iterate over all the ions built into the model by this module and double check their correctness. Looks for tell-tale signs such as negative peaks in the mFo-DFc and anomalous difference maps, abnormal b-factor, and disallowed chemical environments. Prints out a table of bad ions and their information and returns a list of their sites' i_seqs. Parameters ---------- out : file, optional debug : bool, optional segid : str, optional If present, only validate ions with this segid. is segid {0}iNc3 s$|]}|t|fVqdS(N(R(RRh(RH(sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pys asRhRRRRtb_isos2mFo-DFcsmFo-DFcRcRdtratiotBVStVECSUMs9%-15s %-5s %-5s %-8s %-7s %-5s %-5s %-5s %-5s %-6sttitlesValidating new ionstwidthiPR8t-iKRtcS st||ddS(Ntreplace_none_withs---(R(tfsR"((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytffystsuppress_segidiis%.2fs c3 s |]}dj|VqdS(s!!! N(t error_strs(RRU(tprops(sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pys sR( satomsoccsb_isos2mFo-DFcsmFo-DFcsfpsfdpRRR(6R,R#tmodelsRRRRRRRRRDRERsRhRNRRRORSRR-RRRFR$R^RRtflagst use_fp_fdpRcRdR RRRtRBR$RRt peak_2fofct peak_fofcR)RRRfRt at_angstromt valence_sumRt vector_sumRPRwtclose(RHR[RRRRRRRRRDRRTt ion_statusRhRKR^theaderstfmttboxt okay_flagRiRcRdRRBR((RRHsA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt validate_ions<sj   *            $.N($t__name__t __module__t__doc__R,R1RMRWRGRR;RRR<RdtstdoutRRRRRRRRRRRRFR(R)RR6R}RRRR(((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyR!sP C !;   k'  4        9Q   8 C'RcB s eZdZdZdZRS(s Simple wrapper for calling manager.analyze_water with keyword arguments in a parallelized loop. Because the water_result object is not pickle-able, we only return the final ion choice and summary string. cK s||_t||_dS(N(R!Rtkwds(RHR!R((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRMs cC sy|jj||j}tj}|dk r[|jd|d|jjdtn|j }|dkr|d}nd}t |dtst |dd}n|||fSWnt k rdSXdS(NR[RRRRSR[(NNN( R!R}Rt cStringIOtStringIOR,RRR1tgetvalueR6tKeyboardInterrupt(RHRhRR[RR[((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt__call__s     (RRRRMR(((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRs t_validate_ion_wrappercB s eZdZdZdZRS(s Simple wrapper for calling manager.validate_ion with keyword arguments in a parallelized loop. Because the water_result object is not pickle-able, we only return the final ion choice and summary string. cK s||_t||_dS(N(R!RR(RHR!R((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRMs cC s?y&|jj||j}||fSWntk r:dSXdS(N(NN(R!RRRR,(RHRhtcorrect((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRs  (RRRRMR(((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRs t ion_resultcB sHeZdZejjddgZejddZejddZ RS(s Class for validation results for an ion site. This includes information about both the atom modeled at that site as well as its chemical and scattering environment. tchem_envt scatter_envRcC sdS(N((RHR[tprefix((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pytshowscC sdS(N((RHR[R((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt show_briefs( RRRRRt __slots__RdRRR(((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRsR~cB s,eZdZdedZdedZRS(sc Container for storing the results of manager.analyze_water for later display and retrieval. c C st|tdS(N(Rtlocals( RHRKRLRTRURVRWRXRYRZR[R)RS((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRMs c C s~|dkrtj}n|j}t|dkrP|jjddd||jrb|dIJn|jr|dt|dkrdnd IJn|j dk r|d\}}|j t j j kr|jjdt|j d|d |jd tn|d It|IJ|d IJqzt|dkrzg|D]!\}}||jkr8|^q8}t|dkr|d}|d t|IJ|jjdt|d|d t|d IJqMgt|ddD]}t|d^q}|ddj|IJx3|D]+\}}|jjdt|d|qW|d IJqzn*|jtksh|jrz|jjddd||jr|dIJn|jr|dt|j IJ|d IJqz|rmt|jdkrm|dIJt} x{|jD]m} |jjt| rR| r0|dIJt} n|jjdt| d|q|dt| IJqWn|d IJndS(s Prints out a summary of a site's chemical and scattering environment. Parameters ---------- out : file, optional debug : bool, optional iRBtHOHR[s, appears to be nucleotide coordination sites1 Found potential ion%s outside of specified set:itsRRZt confirmeds Probable anion:s( ambigous results, best valence from %sRcS s|dS(Ni((R((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRss ambiguous results, could be %ss, s Probable cation: %ss$ insufficient data to identify atoms possible candidates:s incompatible ligands for %sN(R,RdRRTRERKtshow_propertiesRVRSR[RRRR,tshow_ion_resultsRERZRRYRRPRWRhRXRLRqR1( RHR[RRRRJRGt below_cutoffRURtpossibleR.((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRsr     "         2         N(RRRR,R1RMR(((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyR~s  RcB seZdZed\ZZZZZZ Z Z Z Z ZZZZZZZZZZZZZZZide6de6de6de6de6de 6de 6d e 6d e 6d e 6d e6d e6de6de6de6de6de6de6de6de6de6de6de6de6de6ZdZd0dZ dZ!e"e#dZ$dd1d Z%d!Z&dd"Z'd#Z(d$Z)d0e"d%Z*d&d'd(Z+e,j-d)Z.d0e,j-e#e"d*Z/d+d,d-Z0d0d.Z1e#d/Z2RS(2s Collect physical attributes of an atom, including B, occupancy, and map statistics, and track those which are at odds with its chemical identity. isAbnormally low b-factorsAbnormally high b-factorsAbnormally low occupancysAbnormally high occupancysNo 2mFo-DFc map peaksPeak in mFo-DFc mapsNegative peak in dFo-mFc mapsPeak in the anomalous mapsNo peak in the anomalous maps*Unexpected geometry for coordinating atomss+No distinct geometry for coordinating atomssVECSUM above cutoffsBVS above or below cutoffs$Too few non-water coordinating atomssToo few coordinating atomssToo many coordinating atomssLike charge coordinating likesDisallowed coordinating atomsBad refined f'' valuesDisallowed coordinating residuesBVS far above or below cutoffsBad halide sites Unexpectedly high 2mFo-DFc valuesINo distinct geometry and coordinating another atom with distinct geometrysClose contact to oxygen atomcC s7||_|j|j|_|jjjj|_|jjj |_ |jjj |_ |j |_ |j d|dd}g|_g}x^|D]V}|jdkr|jj|x.|D]}||krPqqW|j|qqWtj|j|_t||_|j|}|j|_|j|_|j|_|j||_|j||_ |j!|_!|j"|_"|j#|_#|j$|_$i|_%i|_&i|_'i|_(i|_)i|_*i|_+i|_,i|_-|j.||_/i|_0|j1||_2||_3dS(NRhRg @RkRl(RkRl(4RhRpRRRRRR(R:RR9Rtstrict_valenceRR1RRRtcount_coordinating_residuestresidue_countsRt geometriesRRRRRRt peak_anomRRlRRRRRRRwtignoredtvectorsRRRGt fpp_expectedtexpected_paramst bad_coordsR(R'RxR)RcR!(RHRhR!tnearby_atoms_unfilteredtnearby_atoms_no_altsRtotherR((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRMUsX                        cC s2|dkr|j}nt|j|dkS(s Calculates whether factors indicate that the atom was correctly identified. Parameters ---------- identity : mmtbx.ions.metal_parameters Returns ------- bool iN(R,RBRERw(RHRB((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRFs cC s-t|j|dko,|j|j|kS(s' Indicates whether the coordinating atoms are of the allowed type (e.g. no N or S atoms coordinating CA, etc.) and residue (e.g. Ser is not an appropriate ligand for ZN). Parameters ---------- identity : mmtbx.ions.metal_parameters Returns ------- bool i(RERtBAD_COORD_RESIDUERw(RHRB((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRqscC su|jt|}|j|p&| }|jt|ot|j|kot|sb|j|kot|j|kot|S(s= More minimal criteria for determining whether a site is chemically compatible, allowing for incomplete coordination shells. Parameters ---------- ion_params : mmtbx.ions.metal_parameters require_anom : bool, optional ignore_valence : bool, optional Returns ------- bool (RwREt"is_compatible_anomalous_scatteringRqtTOO_MANY_COORDtVERY_BAD_VALENCESR(RHR>t require_anomRQRwt anom_allowed((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRssg@c C sd}g}x|jD]}|j|kr4qn|j|kr|jj}|j}|j|j} |jdk r||jkr| |kr|d7}|j | qqqW|S(s< Counts the number of preferred residues coordinating the atom. Used for approximate detection of transition-metal binding sites. Parameters ---------- ion_params : mmtbx.ions.metal_parameters distance : float, optional exclude_atoms : tuple of ... Returns ------- int iiN( R1t atom_nameRRRRtchain_idtresidtallowed_coordinating_residuesR,R( RHR>RRPtn_restresidsRtlabelst other_resnamet other_resid((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRts   cC skd}g}xX|jD]M}|j}||kr|j|krS|d7}n|j|qqW|S(s Counts the number of coordinating atoms within a given radius. Parameters ---------- float Returns ------- int ii(R1tatom_id_no_altlocRR(RHRtn_atomstatom_idsRtother_id((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRus    cC sgd}xZ|jD]O}|jdkr|j|kr_|jtkr\|d7}q\q_qqW|S(s Counts the number of backbone oxygens coordinating a site. Parameters ---------- distance_cutoff : float, optional Returns ------- int iRi(R1RRRRQ(RHRtn_bb_oxR((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRvs cC s|jjdkr@|jdko?|jdk o?|jdkS|jdkrk|jdk oj|jdkS|jd|}||jkr|j|j|kSt S(NRR;g?g@tion(sMGsNA( RRR'R,RRBRxRyRwR1(RHR>RB((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRs cC s|jtkrdn |j}|dkrM|jj|j|jkrMdS|jd|rcdS|j|j |ks|jjdkrdS|j |j |krdSdS(s0 Evaluates whether factors indicate that the atom is lighter, heavier, or isoelectric to what it is currently identified as. Parameters ---------- manager : mmtbx.ions.identify.manager Returns ------- int -1 if lighter, 0 if isoelectronic, and 1 if heavier. RiRBii( RRQRBRR-RRRFt FOFC_PEAKRwt FOFC_HOLE(RHR!RB((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyt atom_weight!s! (cC skddlm}|j|}t}|j|<||j|nX|j8||j?ks|j8||j@krF|r3|j|jAqF|j|jAnt7|j8||j=|jB|R)R?t get_classRBRwRt strict_rulest n_closestt coord_atomsti_pairtcontact1Rtcontact2t n_non_waterRt other_nameRt other_elementtallowedRUto_atomt o_geometryR((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRC>s                    &       ," g333333?g?c C st|}|jj|d}|dkrDt}|j|R)R@RARBRwtfpp_expected_sasakitfpp_expected_henke((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRDs.  ""cC s}|d|jjIJd}|j|j|kr<d}n|j|j|kr[d}n|d|jj|fIJd}|j|j|krd}n|j|j|krd}n|d|jj|fIJd}|j |j|krd}n|j |j|krd}n|d|j |fIJd}|j |j|krSd}n|j |j|krrd}n|d|j|fIJ|jdk rd}|j|j|krd}n|j|j|krd}n|d |j|fIJn|jdk r|d |jIJn|jdk rd}|jd krFd}n|d |j|fIJ|d td|jj|IJn|jdk rydjdd} djdd} |dtg|jD]} | jdkr| ^q| fIJx4|jD])} |d| j| j| fIJqW|jry|dIJx1|jD]#\} }|d| || fIJqLWqyndS(s Show atomic properties that are independent of the suspected identity. Parameters ---------- identity : mmtbx.ions.metal_parameters out : file, optional s%s:Rs <<s %s (%5.3f %s) !!!s' Too few coordinating non-waters !!!s" Too few coordinating atoms !!!s# Too many coordinating atoms !!!s f'' ratio: %6.3f%sN(!RBRRwRRnRR$RR,R*RtunionRoR.RRR+R{RzR!RRRERRGRR(R#RRR'RxRy( RHRBR[RRZRwRtproblemt bad_coordRNRtvector((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRcsT "#")  &#   ig@cC sd}x|jD]~}|j}t|dks|jd krIqn|dd!d kr|ddkr|j|kr|d 7}qqqW||kS(s  Decide whether the atom is coordinating phosphate oxygens from a nucleotide, based on common atom names. Parameters ---------- min_phosphate_oxygen_atoms : int, optional distance_cutoff : float, optional Returns ------- bool iiRitO1tO2tO3tAtBtGi(R(RVRWRX(RYRZR[(R1RRERR(RHtmin_phosphate_oxygen_atomsRtn_phosphate_oxygensRR((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRcs !cC sR|dkr|j}ntjjj|}tjjj|}dj||S(s Covers an atom into a string representing its element and charge. Parameters ---------- ion : iotbx.pdb.hierarchy.atom, optional Returns ------- str s{}{:+}N(R,RRRRfRR RO(RHR RRN((sA/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ions/identify.pyRBs  c C st}|jd sZ      s  g  &