řxTc@ sddlmZddlZddlZddljZddlm Z ddl m Z ddl m Z ddlZdZdZeZdgd Zd ed %siis limit: %.2fti_seqstdistance_idealt distance_cuttweighttslacktlimitttop_outisWARNING: missing atoms!( R tnamet fetch_labelstcountt resseq_as_intti_seqtid_strtdistancetsigmaRSRUt add_proxytadd_nonbonded_exclusion(tacceptor_atomst donor_atomst build_proxiest hbond_countsRPRQt hbond_paramstremove_outlierst use_hydrogensRRR]RSRt donor_labelstacceptor_labelstdonortacceptort acceptor_basethydrogentatomt donor_resseqtdisttsigma_tslack_RT((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt_create_hbond_proxys|                   c &C sd||gkst|j} | dkr6d}n<| dkrKd}n'| dkr`d}n| d| IJdSy|j|j}Wn$tk r}| t|IJdSX|dkst| j|sdSd}|j}x1|jD]#}x|j D] }x|j D]}|j j }|j |}|jdkrXqn|j}t|}d}x|||kr ||}|j }|j}d}d}d} d}!||djr||}"||"}#|#j}$|#jd kr*| d |j|$fIJ|d7}qynL|$||krv| d |j|j|j|#jfIJ|d 7}qyn|#j }%||%djtkr|td |d|%d|d|d|d|d|d| d| d|d|jd|jd| 7}qn|d 7}qyWqWqWqW|S(NRiRiRis( Don't know bonding for helix class %s.itPROs) Proline residue at %s %s - end of helixs- Confusing residue numbering: %s %s -> %s %siR`RaRbRcRPRQRdReRfRRR]RSR(iii(R R@t helix_typet selectiont Exceptiontstrt is_super_sett iselectiontmodelstchainst conformerstatomst extract_i_seqt intersectiontsizetresiduesR2RYRZtresnametidtresidtTrueRrtrestraint_sigmatrestraint_slack(&tparamst pdb_hierarchytselection_cacheRbRRRdRcRPRQReRftmaster_selectionRR)t helix_stepthelix_selectiontet n_proxiest helix_i_seqstmodeltchaint conformert chain_i_seqst both_i_seqsRt n_residuestj_seqtresiduet resi_atomstresseqRiRjRkRltk_seqt bonded_resit bonded_resseqt bonded_atoms((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt"create_helix_hydrogen_bond_proxiess                 %     c C sd||gkst|j} |j} | j| }| j|sOdStd|d|}d}d}xP|t|jkr|j|}| j|j}| j|j }| j|j }|j t dks|j t dkrd|j |j f}t |ntd|d|}d}}t|}t|}|jj|djjdk}|dkrz|dkrztd|d|}td|d|}|dkrN|dkrNxd|krkd|kr|jd ks||d krk|jd krk|jd krB|d 8}|d 8}q|jd kr|d 8}|d 7}qqW|jd kr|d kr| r|d 8}| }n|d kr|r|d 8}| }nx||kr||kr|r ||}||}|d 7}n||}||}|d 7}| }|jjd kr|td |jd|jd|d|d|d|d|d|d| d|d|jd|jd| 7}qqWqw|jd kr"x~||kr|dkr||jjd krv|td ||jd||jd|d|d|d|d|d|d| d|d|jd|jd| 7}n||jjd kr|td ||jd||jd|d|d|d|d|d|d| d|d|jd|jd| 7}n|d 7}|d 8}qWqw| dIJ| d| IJ| d|jIJq| dIJ| d| IJ| d|jIJn)| dIJ| d| IJ| d|jIJ|d7}|} |}|}qsW|S(NiRtstrand_selectionisWrong registration in SHEET record. One of these selections "%s" or "%s" yielded zero or several atoms. Possible reason for it is the presence of insertion codes or alternative conformations for one of these residues.tNRtstart_selectionR1iR0RsR`RaRbRcRPRQRdReRfRRR]RSRs( WARNING: strand direction not defined!s previous: %ss current: %ss3 WARNING: can't find start of bonding for strands!s* WARNING: can't find one or more strands!(R R@tatom_selection_cacheRKRuRxt_get_strand_residuesR2REtbond_start_currenttbond_start_previousRXRRR}tselectRVR=t_find_start_residueR6RRrRR(t sheet_paramsRRbRRRdRcRPRQReRfRRtcachet prev_strandtprev_selectiont prev_residuesRtkt curr_strandtcurr_selectiont curr_startt prev_startt error_msgt curr_residuestitjtlen_prev_residuestlen_curr_residuestcurrent_start_res_is_donort donor_residuetacceptor_residue((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt"create_sheet_hydrogen_bond_proxiesns     *    (                                  cC s^|j}xKt|D]=\}}|jj}|j|jdkr|SqWdS(Nii(Ryt enumerateR}R~RR(RRt start_i_seqsRRt atom_i_seqs((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyRs  cC s>|j}g}x%|jD]}x|jD]}x|jD]}|jj}|j|}|jdkrqEn|j} t | } xMt | D]?\} } | j} || dj t kr|j | qqWt |dkrEPqEqEWt |dkr2Pq2q2Wt |dkrPqqW|S(Ni(RyRzR{R|R}R~RRRR2RRZRR (RRt strand_i_seqststrand_residuesRRRRRRRRRR((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyR s,    cC sn|jj}g|jD] }|^q}g}x3t|D]%\}}|jdkrt|d|dIJqAnd|j} || } || d} || d} |jdkrd} ntj|j} t j j j d|dd |dd | j d | jd | jd | jd| j d| jd| jd| jd| ddd| j }|j|qAW|S(NsEmpty helix at serial %d.is0(%s) and (name N) and (altloc 'A' or altloc ' ')iiRitserialthelix_idt start_resnameR R#R$t end_resnameR!R%R&R)tcommentR tlength(RRytatoms_with_labelsRRuR RtR(tindextiotbxtpdbtsecondary_structuret pdb_helixRtchain_idRticodeRR (RR RtiseltaR}t pdb_helicesRt helix_paramstsele_strt amide_iselt start_atomtend_atomR)t current_helix((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pytrestraint_groups_as_pdb_helices&s>             cC sM|jj}g|jD] }|^q}g}xt|D]\}}tj|} |jdkrz|d| IJntj j j d| ddt |j dgdg} td|d|jd |d |d d} | j| | jdd } x:|j D]/} td|d| jd |d |d | j}| j|| | j}| | j}||}||}|jd ks|jd kr| jdqn||d }||d }tdtd|jd|jd|jd|jd|jddd|jd|jd|j }| j|qW|j| qAW|S(Ns Missing first strand in sheet %stsheet_idt n_strandsiR3R5RRuR}RR6s.(%s) and name N and (altloc 'A' or altloc ' ')iR<t cur_resnameR;R7R8R?t prev_resnametOR>R9R:(RRyRRtstringt uppercaseRKR RRRt pdb_sheetR2REt_strand_group_as_pdb_strandt add_strandtadd_registrationRuR6RRRRt donor_nameRRRRRVR (RRRRRR}t pdb_sheetsRRRt current_sheetRKt base_seleREt pdb_strandts1ts2t reg_curr_iselt reg_prev_iselt reg_curr_atomt reg_prev_atomRF((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pytrestraint_groups_as_pdb_sheetsHsd               $         c C s|dks|dkr!d}n*|dkr6d}n|dkrKd}n|d|}||d}||d}tdtd tdd |jd |jd |jd |jd|jd|jd|jd|jd| } | S(NRiR1iR0is.(%s) and name N and (altloc 'A' or altloc ' ')Rt strand_idRR R#R$RR!R%R&R6(R RRRRRRR( RRuR}RR6t int_senset strand_iselRRR((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyR|s.                tfind_helices_simplecB seeZdZdZdZdZdZdZdZdZ dZ e j d Z RS( s Identify helical regions, defined as any three or more contiguous residues with phi and psi within specified limits: -120 < phi < -20 -80 < psi < -10 This is much more tolerant of distorted models than KSDSSP, but will still miss helices in poor structures. cC s)||_g|_g|_|jdS(N(Rt_helicest_current_helixtrun(RR((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt__init__s   cC s8t|jdkr+|jj|jng|_dS(Ni(R2RRR (R((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt process_breakscC sft|jdkrP|jd\}}||krb|jj||fqbn||fg|_dS(Nii(R2RR (RRRt last_chaint last_resseq((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt push_backs  c C s.ddl}ddl}|jj}|jj}|j}g}x|jjD]}x|jD]}|j d} | j } xt t | dD]}} | | } || j dkrd} | dkr| | d} n| | d}| j}|j}| dk r||dksI||kr| j|jkr| j}||dks||kr| j| jkr|jjd| d| d|}|jddkrqn|jjd|d |\}}t||r |j|j|qq qqn|jqqWqhWqUWdS( Niiitcommon_amino_acidtprev_resRtnext_resROt sites_cart(t mmtbx.rotamert iotbx.pdbRtcommon_residue_names_get_classRR}t extract_xyzRzR{R|RtrangeR2RR RYRtrotamertget_phi_psi_indicesRXtphi_psi_from_sitestis_approximately_helicalRRR(RtmmtbxRt get_classR}RRRRt main_confRti_restresidue1tresidue0tresidue2tresseq1tresseq2tresseq0tphi_psi_i_seqstphitpsi((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyRsP                cC sdg}xW|jD]L}|dd}d|dd|dd|ddf}|j|qW|S(Nischain '%s' and resseq %d:%dii(RR (Rtatom_selectionsRRRu((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pytbuild_selectionsscC sg}x+|jD]}d|}|j|qWt|dkrtjjt}tjjdj|}|jd|SdS(Nshelix { selection = "%s" }is tsource( RR R2Rtphiltparsethelix_group_params_strRtfetchR (RRRut helix_strt master_philR((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pytas_restraint_group_phils cC s&|j}|dk r"|jSdS(N(R R textract(RR((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyRs   cC sdS(N((R((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pytas_pdb_recordsscC sbt|jdkr"|dIJn|dt|jIJx"|jD]}|d|IJqFWdS(NisNo recognizable helices.s%d helix-like regions found:s %s(R2RR(RtoutRu((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pytshows  (RRt__doc__RRRRRR RR"RRR$(((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyRs    +  cC s@d|kodknr<d|ko3dknr<tStS(Niiii(RtFalse(RR((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyR s8g@cC s|dkst|j|jks0t|j|jksNt|j}|}d} x|D]} x{|D]s} || \} } }|| \}}}t|| d|| d||d}||krz| tf} qzqzWqmW| S(NRi(NN(R@RRR Rtacceptor_i_seq(R}tprev_strand_donorstprev_strand_acceptorstcurr_strand_donorstcurr_strand_acceptorsR6tmax_distance_cutoffRtmin_distt best_pairt donor_i_seqtacceptor_j_seqtx1ty1tz1tx2ty2tz2Ro((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyt_find_strand_bonding_start s   , (+t __future__Rtiotbx.pdb.secondary_structureRt iotbx.philtlibtbx.object_oriented_patternstobject_oriented_patternstoopt libtbx.utilsRtlibtbxRtmathRRRtsheet_group_params_strRR"R(tinjectorRRt annotationRRRRR/R RRrRRRRRRRtobjectRR R7(((sP/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/secondary_structure/proteins.pyts>   %    %%%UO\ "4 l