xTc@ sddlmZddlmZddlmZddlmZddlm Z ddl m Z ddl Z de fd YZd e fd YZd e fd YZddeddZdZdZdZdZdZddZdZdddZdddZdddZdZdZd Zd!d"dd#Z d$Z!d%Z"d&Z#ddd'Z$dS((i(tdivision(t eigensystem(tflex(t superpose(tSorry(tmatrixNtNCS_groups_containercB seZdZRS(cC s(g|_g|_g|_g|_dS(sq A Container for ncs groups Note that the first copy is the master ncs Attributes: iselections (list of flex.size_t):selection array for the complete ASU residue_index_list (list): list of list of matching residues indices copies (list of lists):List of lists of the ncs copies chain IDs transforms (list of transform objects): N(t iselectionstresidue_index_listtcopiest transforms(tself((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyt__init__ s   (t__name__t __module__R (((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR st TransformcB s#eZdddddddZRS(icC s@||_||_||_t||_||_||_dS(s Basic transformation properties Args: rotation : Rotation matrix object translation: Translation matrix object serial_num : (int) Transform serial number coordinates_present: equals 1 when coordinates are presents in PDB file ncs_group_id : (int) The group ID of the group containing this transform rmsd (float): RMS distance between ncs copies N(trttt serial_numtbooltcoordinates_presentt ncs_group_idtrmsd(R trotationt translationRRRR((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR s     N(R RtNoneR (((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyRs t Score_recordcB s&eZdddddZdZRS(i iicC s:||_d|_d|_||_||_||_dS(s score object used when aligning sequences Attributes: score (int) consecutive_matches (int): num of consecutive matches in current segment match_count (int): number of matching residues min_contig_length (int): segments < min_contig_length rejected gap_penalty (int): penalty for gaps origin (tuple): (row,col) of the matrix cell we from the previous alignment step. Used to trace back optimal alignment iN(tscoretconsecutive_matchest match_counttmin_contig_lengtht gap_penaltytorigin(R RRRR ((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR 8s      cC s\t|j|}|j|krO|j|j|j8_|j|j8_nd|_dS(s Args: seq_length (int): length of shortest of the compared lists When there is a gap in the matching, reset the consecutive_matches and adjust the score if consecutive_matches < min_contig_length iN(tminRRRRR(R t seq_lengtht min_matches((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pytreset_match_countingNs (ii(R RR R$(((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR6si g?g@c C smtd|d|d|}t|||}|rHt|\}}nt|\}}t||}|S(s Find NCS relation in hierarchy Args: ph (object): hierarchy min_contig_length (int): minimum length of matching chain segments min_fraction_domain (float): Threshold for similarity between chains similarity define as: (number of matching res) / (number of res in longer chain) use_minimal_master_ncs (bool): use maximal or minimal common chains in master ncs groups rmsd_eps (float): limit of rms difference chains when aligned together Return: groups_list (list of NCS_groups_container objects) group_dict (dict): keys: tuple of master chain IDs values: NCS_groups_container objects tphRtmin_fraction_domain(tsearch_ncs_relationstclean_chain_matchingtminimal_master_ncs_groupingtminimal_ncs_operators_groupingtbuild_group_dict( R%RR&tuse_minimal_master_ncstrmsd_epstchain_match_listt match_dictttransform_to_groupt group_dict((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pytfind_ncs_in_hierarchy\s c8C st|}d|D}t|}i}i}t}d}xj|D]b\}}|||f\} } } } } }}||krqJn||krqJnt| ||\}}|rD|||dkrd}||}}qQ||}}|j||ft| |\} }| | | | | ||g|||fs iiN(tsortedtsettfind_same_transformRtpoptinverse_transformthas_keytappendtupdatet iteritemstindextlentsortRt intersectiontaddtkeys(8R/tsorted_chain_groups_keyst chain_idsR0tcopy_to_transformtchains_in_groupsttr_snt master_idtcopy_idtsel_1tsel_2tres_1tres_2RRRttr_numt is_transposet temp_mastert temp_copyt copies_idsttr_listttrtktvtmastersR t common_idstch_idtitmultiple_tr_numtsingles_tr_numtm_tr_numtm_mtm_cts_tr_numts_mts_ctt1tt2tt3tt4tt5t master_sizetchain_left_to_addttransform_to_usetchains_in_mastertchains_in_copiestntrotttransttest1ttest2ttest3RDR4((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR*s   %   (    !       ""  5     "    cC st|}d|D}t|}i}t}d}x|D]\}}|||f\}} } } } } }||ks||kr|j||fdqDnt| | |\}}|rQ|||dkrd}||}}q^||}}|j||fdt| | \} } | || | | | |g|||f4s iiN( R6R7R9RR8R:RCR<R>R@RARRD(R/RERFR0RHRIRJRKRLRMRNRORRRRPRQRRRSRWRXRjRkRlRoRVttr_sns((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR)sV   % (   #5        cC si}d}d}x|jD]\}}|\}}\} } t|} g| D]} |j| ^qS} g| D]}||^qu}t| }t| ||\}}|j|r2t||||jd}xt|D]\}}|\}}}t |t |}t |t |}||krt ||j|||\}}|||<|||_qqW|d7}|| d|df\}}}}} } }t d| d| d|dt d|d|}||jj|||jj|||jj|||jj|q|d7}|d7}|| d|df\}}}}} } }t}t dtjdddddddddg dtjdddgd|dt d|dd}|jj||jj||jj| |jj||d7}t d| d| d|dt d|d|}|jj||jj||jj||jj||||R6R?ttupletget_iselectionR;tzipRt enumerateR7R!tupdate_selectionsRtTrueR<RR R RRtsqrtcol(R0R/R1tgroup_idRIt_RXRYR RRtsorted_masterstxtsort_mapR\R4t m_sel_listt c_sel_listt zipped_paramtrectm_seltc_seltcurrent_mastertcurrent_master_sett m_sel_settupdated_iselectiontres_l_mtres_l_cRRVt new_ncs_group((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR+us~ "    -   - *   cC s,|d|}t|t|}t|}t||}||}||} |rg} xw|D]f} | |} t| } t| | }t||} tj| | } | | |<| j| qgWn|} | r"t|} t|| }t|| }tj|| }n| |fS(s| Update the total master and copies selection with the smallest common selection, among all ncs copies Args: iselections (list): list of flex.size_t iselection for all NCS copies m_sel,c_sel (flex.size_t): iselection for master and a single copy i (int): chain number in iselections Returns: iselections : updated iselections c_sel : updated c_sel i(R7R!tlistRtsize_tR<(RRRR\RRt master_offsetRtremove_from_old_ncs_copiestremove_from_new_copytnew_iselectionst isel_listtiseltoffsettisel_sett c_sel_set((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyRzs.        c C sig}g}xPt||D]?}||\}}}}}}}|j||j|qW||fS(s] Combine iselection of all chains in NCS master and NCS copy to a single iselelction Args: sorted_masters (list): list of master chain IDs copies (list): list of copies chain IDs match_dict(dict): key:(chains_id_1,chains_id_2) val:[selection_1,selection_2,rot,trans,rmsd] chain_ID (str), selection_1/2 (flex.size_t) residue_index_list_1/2 (list): matching residues indices rot (matrix obj): rotation matrix tran (matrix obj): translation vector rmsd (float) Returns: m_sel,c_sel (list of flex.size_t): list of selected atoms per chain (RxR<( RR R/RRR4RLRMR((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyRws ic sHg|D]}|ddkr|^q}i}i}x|D]}|\}} } } } } }t|||| |\}}|j| jj}|j| jj}tjd|d|}|j}|j}|j|j }t |d}| | | | |||g||| fCs ( tupdate_best_matches_dicttselecttatomst extract_xyzRtleast_squares_fitRRtrms_differencetother_sites_best_fittroundR>(R.R%R-Rt match_listt best_matchesR/tmatchtch_a_idtch_b_idtsel_atsel_bt res_list_1t res_list_2t similarityRt ref_sitest lsq_fit_objRRR((R-sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR(s&)     )c C sbt}x"||gD]}|j|rg}||dd}||krx]||D]Q\} \} } | |dkr|j| | | fgqZ|j| | fqZW|j|||fgn9||dkr|jd|||fgn ||}|||thasattrRRtis_same_transform( RRR RRRQRPRWRXtrrtttt is_the_same((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR8ss  $c! C s9|j}g}|jd}td|jD}t|}t|}|j}xt|dD]} || j} |jd| } |j | } || } x|t| d|D]g}||j}|jd|}|j |}||}t | \}}t |\}}g|D]\}}|^q0}g|D]\}}|^qO}t d|d|d|d|\}}}t | |||| |\}}}}|r|j }|j }|t| j 7}|t|j 7}| ||||||g} |j| qqWqsW|S( s Search for NCS relations between chains or parts of chains, in a protein hierarchy Args: ph (object): hierarchy min_contig_length (int): segments < min_contig_length rejected min_fraction_domain (float): Threshold for similarity between chains. similarity define as: (number of matching res) / (number of res in longer chain) Returns: chain_match_list (list): list of [chain_ID_1,chain_ID_2,sel_1,sel_2,res_sel_m, res_sel_c,similarity] chain_ID (str), sel_1 (flex.bool), sel_1 (flex.bool), res_sel_m/c (lists): indices of the aligned components similarity (float): similarity between chains We use selection_1 and selection_2 because in some PDB records a residue might contain side chain in one chain but not in another icS sh|]}|jqS((tid(R3R((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pys s ischain tseq_atseq_bRR&(tatom_selection_cachetmodelsRtchainsR6R@txrangeRt selectionRtget_residue_sequencet res_alignmenttget_matching_atomst iselectionR!R<(!R%RR&tcacheR.tmodelRFtn_chainst ph_chainsR\t master_ch_idt m_selectiontm_phtm_chaintjt copy_ch_idt c_selectiontc_phtc_chaint chain_id_mtres_mt chain_id_ctres_cRtytseq_mtseq_ct res_sel_mt res_sel_cRtsel_mtsel_cR((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR'sH         $   c C s t|}t|}|dks0|dkr=ggdfSd}d||}td|||}|} td|d|d| d|d|} xtd|dD]} xtd|dD]} || dj|| djk} | | d| dfj|| }| | d| fj|}| | | dfj|}t|||}td|d || | | f<||kr| | d| dfjd| | | f_| d| df| | | f_ | | d| dfj d| | | f_ q||kr$| d| }}n| | d}}||f| | | f_ | ||fj | | | f_ | ||fj| | | f_| | | fj t ||qWqW| | | fj t ||t d | d|d|d |\}}}|||fS( sz Align two chains hierarchies. Penalize misalignment, gaps, when contiguous section is less than min_contig_length and when the Do not give any points for alignment. (score only change when "bad" things happens) Stop if the alignment is less than the "similarity_threshold" After alignment verify that aligned parts are >= similarity_threshold Args: seq_a, seq_b (lists of str): list of characters to compare min_fraction_domain (float): min percent of similarity between hierarchies similarity define as: (number of matching res) / (number of res in longer chain) min_contig_length (int): domain < min_contig_length rejected Returns: aligned_sel_a (list): the indices of the aligned components of seq_a aligned_sel_b (list): the indices of the aligned components of seq_b similarity (float): actual similarity between hierarchies iiitrowR}t max_scoreRRRtRR&(R@tinttbuild_score_matrixRtlowerRtmaxRRR RR$R!tget_matching_res_indices(RRRR&RRRtmisalign_penaltyt max_mis_alignRRRR\t not_alignedts1ts2ts3Rti_temptj_tempt aligned_sel_at aligned_sel_bR((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyRsD   (# ,!/   ( $c C s|}|}g}g}|j||fsLtj|tj|dfS|||fjt||}||krtj|tj|dfSx|dkr|dkr|||fj|d|dfkr|j|d|j|d|d8}|d8}q|||fj|d|fkrB|d8}q|||fj||dfkrr|d8}qtdqW|j|jt |t |kst |||fS(s Trace back best solution and collect the indices of matching pairs Args: R (list of lists of int): the score matrix row (int): number of rows col (int): number of columns min_fraction_domain (float): min percent of similarity between hierarchies similarity define as: (number of matching res) / (number of res in longer chain) Returns: sel_a (list): matching indices sequence a sel_b (list): matching indices sequence b similarity (float): actual similarity between hierarchies iisSequence alignment error( R;RRRRR R<RtreverseR@tAssertionError( RRR}R&tj_maxti_maxRRR((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyRs0  '  # #   cC s7|jj}|jj}|jj}|jj} tjtg|} tjtg| } g|jD]} | j^q{} g|jD]} | j^q}g}g}x`t | |||D]I\}}}}|d|}|d|}|j |}|j |}t |jj }t |jj }t d|d|dd\}}}tj|t|j}tj|t|j}|jdkr|j||j|ntj||}tj| |}| |B} | |B} qW| | ||fS(s Get selection of matching chains, match residues atoms We keep only residues with continuous matching atoms Args: ph_a (hierarchy): first chain ph_b (hierarchy): second chain res_num_a/b (list of int): indices of matching residues numbers, ph_a/b chain_a/b (chain hierarchy): chains hierarchies of chains a and b Returns: sel_a, sel_b (flex.bool): matching atoms selection res_num_a/b (list of int): updated res_num_a/b sresid RRRidi(RRRtsizeRRRtresiduestresidRxRRt extract_nameRRR!RR<(tph_atph_bt res_num_at res_num_btchain_atchain_bt atom_cache_at atom_cache_btl_atl_bRRRt res_ids_at res_ids_btres_num_a_updatedtres_num_b_updatedtresid_atresid_btnatnbtsatsbtres_atres_bt atoms_names_at atoms_names_btatoms_atatoms_bR((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyRIs<%%(   cC sg|jD]}|j^q }g|jD]}|j^q/}gt||D]*\}}|jdkrT||f^qT}|j}|djj}||fS(s Get a list of pairs (residue name, residue number) and the chain ID Exclude water Args: chain_ph (iotbx_pdb_hierarchy_ext): hierarchy of a single chain Returns: chain_id (str) res_seq (list) thohi( t atom_groupstresnametresidue_groupstresseqRxRRtchainR(tchain_phRtres_listt resid_listRtres_seqRtchain_id((sB/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/ncs/ncs_search.pyR{s ""@ iic C si}xUt|dD]C}|||}td|d|d|ddf||dfsF ' $ Y a )  * -! A @ 2 2