xTc@ s ddlmZddlmZddlmZddlZddlmZmZddlm Z m Z ddl Z ej j dZdZd Zd Zd Zd Zd efdYZdefdYZeee dee jd dZdZdZdS(i(tdivision(teasy_mp(t str_utilsN(tSorrytnull_out(tadopt_init_argstAutosx similarity_matrix = blosum50 dayhoff *identity .type = choice min_allowable_identity = 0.5 .type = float iitUtKcC s]t|dkrdSt|dkr0|dSddlm}|j|}|jSdS(Niii(tflex(tlentNonetscitbx.array_familyR t vec3_doubletmean(tsitesR tv((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytget_mean_coordinatestchaincB seZdZdddZdZdZdZdZdZ dZ dZ d Z d Z d Zed Zed ZRS(sN Stores information on a protein or nucleic acid chain, its alignment to the target sequence, and the coordinates of each residue. For command-line use, much of this is irrelevant. In the PHENIX GUI, much of this data is fed to the sequence/alignment viewer (wxtbx.sequence_view), which controls the graphics window(s). cC st|t|ttgks(tt|t|ksFtd|_d|_d|_ d|_ d|_ d|_ d|_ d|_g|_g|_g|_g|_g|_d|_g|_dS(Ngi(RtlocalstPROTEINt NUCLEIC_ACIDtAssertionErrorR R t alignmentt sequence_namet sequence_idtidentityt n_missingtn_missing_startt n_missing_endtn_gapstextratunknowntmismatcht actual_codet_xyzt_tablet _flag_indices(tselftchain_idtsequencetresidst chain_typetsec_strtresnames((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt__init__'s$              cC szt|jt|jkr3t|jdks9t||_||_||_|jdk rut |j}nt dt|j }d|_|jd krd|_n|dk r|j ddg|_ nd}}t }d}t|j} x|| kr|j|} |j|} d} | dkrf|j|} |j||7_|d7}n|jd7_||j} | dkr| d kr|jd7_| r|jdks|j| dkr|jd7_q|dkr|jd7_qqn| dkr|r| t|jkr|jj| d|j| d|jdk rw|jj| dn|d7}qq| d kr|jd7_|jd7_qn| d krt}| dkr|jj| |jj|q| d kr| | kr| tkr+|jtksF| tkrY|jtkrY|jj| n |jj| |jj| |jj|qn| }|d7}qW| d}|j }x|j|dkrQ||j}||dkrD|jdks#|t|jks#|j|dkrD|j!d7_!|jd7_n|d8}qWt|jt|jksvtdS(Nit-ts (unnamed)t skip_charstXit (NR/(R1R.(R.(R1R.(R1R.(R1R.("R tatbRRRRR+R tlistR(tcalculate_sequence_identityRtFalseR)RRR,RtinserttTrueRtappendR%tUNK_AAR*RtUNK_NARR R!R"tmatchesR(R&RRRt raw_sec_strti_alnti_residt chain_startedt prev_chartalignment_lengtht symbol_pdbt symbol_seqtresidtj_residR=((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt set_alignment<s9               "        cC s|j|jkstg|_d}x^|jD]P}|j}x1|j||kry|jjd |d7}qIWd }||j krd}n0||j krd}n||j krd}nd }x[|j D]1}|jdks|jdkr|j}PqqWd|GH|j dj}|jj||d k rz|jj|j||d |j|f|gn|d7}q4Wx3|t|jkr|jjd |d7}qWt|jt|jkstd S( s Collect the coordinate of the central atom (CA or P) in each residue, padding the array with None so it matches the sequence and resid arrays. iisnot in sequencesmismatch to sequencesspecial residues CA s P s(WARNING: can't find center of residue %sschain '%s' and resid %sN(R'tidRR$tresidue_groupsRFR)R#R:R RR!R tatomstnametxyzR (R&t pdb_chaintkt residue_groupRFt res_classRMtatom((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytextract_coordinatess>        cc s|j|jkstd}xX|jD]J}|j}x&|j||kre|d7}dVq@W|V|d7}q+Wx(|t|jkr|d7}dVq|WdS(Nii(R'RIRRJRFR)R R (R&RNRORPRF((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytiter_residue_groupss    cC st|j||_dS(N(R5RTRJ(R&RN((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytextract_residue_groupsscC s|jS(sUsed in PHENIX validation GUI(R$(R&((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytget_outliers_tablescC s|jS(s:Used for wxtbx.sequence_view to highlight mismatches, etc.(R%(R&((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytget_highlighted_residuesscC st|jdkstd}g}xlt|jjD]X\}}||krVPn#||kry|j|j|n|dkr:|d7}q:q:W|S(NiR.i(R R#Rt enumerateRR3R:(R&ti1ti2RORtjR3((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt#get_coordinates_for_alignment_ranges   cO s|j||}t|S(N(R\R(R&targstkwdsR((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt'get_mean_coordinate_for_alignment_rangescC s:g}x-|D]%\}}|j|j||q W|S(N(textendR\(R&trangesRRYRZ((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt$get_coordinates_for_alignment_rangesscO s|j||}t|S(N(RbR(R&R]R^R((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt(get_mean_coordinate_for_alignment_rangesscC s<|r"|jj|jj|jgS|jj|jjgSdS(N(RR3R4R+(R&tinclude_sec_str((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt get_alignmentsc  sfd}d|jIJ|jdkr<dIJncd|jIJd|jdIJ|jdkrd|j|j|jfIJn|jdkrd |jIJnt |j dkrd t |j IJ||j nt |j dkr/d t |j IJ||j nt |j dkrkd t |j IJ||j n|r|jj d dddddddtndS(Nc st|dkstd}djg|D]}|j^q+}tj|djd}d|dIJx |dD]}d|IJq~WdS(NiR2i<s s residue IDs: %sis %s(R RtjointstripRtwordwraptsplit(R)twRFtlinetlines(tout(sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt print_residss(s Chain '%s':s& No appropriate sequence match found!s best matching sequence: %ss sequence identity: %.2f%%idis? %d residue(s) missing from PDB chain (%d at start, %d at end)s %d gap(s) in chains %d mismatches to sequences# %d residues not found in sequences %d residues of unknown typeRmt block_sizei<ttop_namesPDB filet bottom_nameR(t show_ruler(R'RR RRRRRRR R!RR t pretty_printR7(R&RmtverboseRn((RmsG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt show_summarys2 #N(t__name__t __module__t__doc__R R-RHRSRTRURVRWR\R_RbRcR7ReR9Ru(((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyRs  L &        t validationcB sbeZddeeeeddZdZdZdZddZ dZ ddZ RS( ic 'C s^t|dkstx#|D]} | jdkstqW|dkrVtj}n|dkrqtj}nd|_d|_ d|_ g|_ | |_ ||_ dgt||_xtdt|D]m} || } xZtdt|D]C} | | krPq|| }| j|jkr| |j| ||krd$d%|t|fd&d'} |t kr;|d4k r;|r;|| IJ|d(IJ|d)IJ|d*IJ|d+|IJ|SqL||kr-d,d-|t|fd&d'} qL|d.IJdSnd/d0|} | d4k r|r|d1| IJ|d(IJ|d2IJdS| | n td3d4S(5s Utility function for reconciling the contents of the sequence file, the chains in the model, and the ASU. Returns the number of copies of the sequence file to tell Phaser are present, or raises an error if this is either ambiguous or in conflict with the search model multiplicity. This is intended to allow the user to specify any combination of inputs - for instance, given a tetrameric search model, 2 copies in the ASU, and a monomer sequence file, the ASU contains 8 copies of the sequence(s). s=Guessing the number of copies of the sequence file in the ASURzRR}RiRRSRg333333?cS st|ddddS(Ns5 (Add the parameter force_accept_composition=True to sDdisable this error message and continue assuming 1 copy of sequence sfile.)(R(tmsg((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt raise_sorry5s is>%d chain(s) not found in sequence file. If the sequence does sEnot accurately represent the contents of the crystal after adjusting s=for copy number, molecular replacement and building may fail.s WARNING: %sis8WARNING: %d sequence(s) not found in model. This is nots;usually a problem, but it may indicate an incomplete model.s!WARNING: assuming 1 copy of models%Must have at least one copy of model.s8The sequence file provided does not appear to match the s search model.s6The sequence file provided does not map evenly to the sCsearch model; this usually means an error in the sequence file (or s>model) such as accidental duplication or deletion of a chain. s [counts: %s]cS st||kS(N(tint(tx((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyt is_integer^sg?s6Assuming %d copies of sequence file (as well as model)s+Assuming %d copies of sequence file presents4The search model does not appear to contain a round s5number of copies of the sequence; this usually means s<that your sequence file has too many or too few sequences. s [freq=%g]s8The number of copies of the model expected is less than s;the number indicated by the sequence file (%d versus %d). s=This may mean that either the sequence file or the predicted snumber of copies is wrong.R/s1Since the number of copies was guessed by Xtriages-based on the sequence file, it will be scaleds-by %g to be appropriate for the search model.s7The number of copies of the model expected exceeds the s5number indicated by the sequence file (%d versus %d).s1Assuming only one copy of sequence file contents.s5The sequence file does not appear to contain a round s#number of copies of the model (%g).s WARNING: s6Ambiguous input, defaulting to 1 copy of sequence files%No error but frequency not determinedN(RyR9R7RR RtsettremovetcountR RRR5RRt RuntimeError(RRztforce_accept_compositiontcopies_from_xtriagetcopies_from_userRt(assume_xtriage_copies_from_sequence_fileRmRRRRterrorRt unique_countstfreqt model_copiesRtseq_freqt inverse_freq((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pytget_sequence_n_copiess                                           cK sddlm}|j|dtdt}|jdkrItdn|j|dtdt}|jdkrtdn|jj|d <|j|d              (t __future__RtlibtbxRRt libtbx.philt libtbx.utilsRRRRRtphiltparseRRRR;R<RtobjectRRyR7R RRRR(((sG/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/validation/sequence.pyts2