ó Ø™xTc@ sÒddlmZddlZddlZddlZddlmZddlmZddl m Z ddl m Z ddl m Z ddlmZdd lmZddlmZddlZejjd d eƒZd efd „ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZd„Zd„Zdd„Zd„Zd„Zd„Zd„Z dd„Z!dd„Z"d „Z#d!„Z$d"„Z%d#„Z&id$„Z'd%„Z(dd&„Z)ddej*d'„Z+ddej*d(„Z,dd)ddej*d*„Z-dd)ddej*d+„Z.d,ej*d-„Z/ej*d.„Z0dej*d/„Z1ej*d0„Z2ej*d1„Z3dd)ddej*d2„Z4ddd3„Z5ddd4„Z6dd5„Z7d6„Z8dS(7iÿÿÿÿ(tdivisionN(tpdb(t file_reader(t cablam_res(t cablam_math(t NDimTable(t easy_pickle(teasy_runsÖ cablam_validate { pdb_infile = None .type = path .help = '''input PDB file or dirpath''' output = *text kin ca_kin full_kin markup_no_ribbons points records records_plus_pdb oneline .type = choice .help = '''choose output type, =text for comma-separated outlier summary =kin print markup kinemage to screen =ca_kin print CA geo markup kinemage to screen =full_kin open in King a multi-crit-type kinemage with markup =markup_no_ribbons print allowed, outlier, and ca geom outlier markup kin =points print cablam-space points to sys.stdout, in kinemage dotlist format =records print HELIX and SHEET-style records based on CaBLAM analysis =records_plus_pdb print HELIX and SHHET-style records as part of a pdb file =oneline print one-line validation to sys.stdout: percent of cablam outliers ''' outlier_cutoff = 0.05 .type = float .help = '''sets the contour level for detecting outliers, e.g. outlier_cutoff=0.01 for accepting the top 99%, defaults to 0.05''' check_prolines = False .type = bool .help = '''test of a function to differentiate cis vs trans prolines''' help = False .type = bool .help = '''help and data interpretation messages''' } tprocess_includestcablam_validationcB seZdddd„ZRS(cC s‚||_||_||_d|_d|_d|_d|_d|_d|_ d|_ |jru|jj |_ n d|_ dS(Nt( tresiduet outlier_leveltca_geom_outlier_leveltNonet loose_alphat regular_alphat loose_betat regular_betatloose_threetentregular_threetent suggestiontrg(tselfR R R ((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyt__init__ks           N(t__name__t __module__RR(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR jst motif_valuecB seZd„ZRS(cC s:d|_d|_d|_d|_d|_d|_dS(N(RRRRRRR(R((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR}s      (RRR(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR|st motif_guesscB seZd„ZRS(cC s:t|_t|_t|_t|_t|_t|_dS(N(tFalseRRRRRR(R((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR†s      (RRR(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR…st motif_chunkcB s_eZejd„Zdejd„Zdejd„Zddddejd„Zd„ZRS(cC s„|jdkr"|jd|ƒn^|jdkrD|jd|ƒn<|jdkrf|jd|ƒntjjdƒtjƒdS(Nthelixtwritetotthreetentsheets9 tried to print a record that was not 'helix' or 'sheet' (t motif_typetprint_helix_recordtprint_threeten_recordtprint_sheet_recordtsyststderrtwritetexit(RR ((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyt print_records icC s¬|jjƒ}|jjƒ}|jj|jjd}|dkr¨|jdd|dd|d|d d|dd|d d|ddd|d ƒndS( NiisHELIX s%3it is! 1 s%5is (t motif_starttid_with_resnamet motif_endtresnumR)(Rt helix_numR tmotif_start_idt motif_end_idt motif_len((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR$›s ScC s¬|jjƒ}|jjƒ}|jj|jjd}|dkr¨|jdd|dd|d|d d|dd|d d|ddd|dƒndS( NisHELIX s%3iR,is! 5 s%5is (R-R.R/R0R)(RR1R R2R3R4((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR%¦s StUicC sg|jjƒ}|jjƒ}|jdd|dd|d|d|d|d|dƒdS(NsSHEET s%3iR,s%3ss%2is (R-R.R/R)(Rtsheet_idt strand_numt strand_countt strand_senseR R2R3((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR&±scC sd|_d|_d|_dS(N(RR-R/R#(R((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyR¹s  ( RRR'tstdoutR+R$R%R&R(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pyRŽs    cC stjjdƒdS(Nsv phenix.cablam_validate file.pdb [options ...] Options: pdb_infile=filename input PDB file some output formats support a dir of files in this field outlier_cutoff=0.05 sets the contour level for detecting outliers, e.g. outlier_cutoff=0.01 for accepting the top 99%, defaults to 0.05 output= kin : prints markup kinemage for a single outlier cutoff to screen text : prints machine-readable columnated and comma- separated data to screen (default output) points : prints cablam-space points to screen, in kinemage dotlist format records : prints HELIX and SHEET-style records to screen, based on CaBLAM secondary structure analysis records_plus_pdb : prints HELIX and SHEET records and the associated pdb file to screen ca_kin : prints a kinemage with ca-contour outliers marked full_kin : opens a phenix.king window with multicrit style markup for the submitted structure. Also saves a .pdb with HELIX and SHEET records and a .kin with other markup to working dir markup_no_ribbons : prints 'allowed', 'outlier', and 'ca geom outlier' kinemage markup to screen oneline : prints oneline-style output to screen. Supports printing for multiple files help=False prints this usage text, plus notes on data interpretation to screen Example: phenix.cablam_validate file.pdb output=full_kin -------------------------------------------------------------------------------- (R'R(R)(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pytusageÂs *cC stjjdƒdS(Ns€ cablam_validate data interpretation: Text: Text output is provided in comma-separated format: residue,contour_level,loose_alpha,regular_alpha,loose_beta,regular_beta 'residue' is a residue identifier formatted as pdb columns 18 to 27 (1 index) 'contour_level' is the 3D outlier contour at which this residue was found. Smaller values are worse outliers. Residues with values <=0.05 are worth concern. 'loose_alpha' is a 2D percentile contour for this residue's similarity to alpha helix. A high value indicates some amount of helix character, but not necessarily the presence of a regular helix. >0.001 is meaningful, >0.01 is likely 'regular_alpha' is a 2D percentile contour for this residue's similarity to regular alpha helix. The contours for regular helix are very tight. >0.00001 is meaningful, >0.0001 is likely 'loose_beta' is a 2D percentile contour for this residue's similarity to beta strand. A high value indicates some amount of strand character, but not necessarily the presence of a regular strand. >0.01 is meaningful, >0.1 is likely 'regular_beta' is a 2D percentile contour for this residue's similarity to regular beta sheet. Beta structure contours are generally more permissive than helix contours, and the contour values should not be judged on the same scale. >0.001 is meaningful, >0.01 is likely ('meaningful' and 'likely' scores are subject to change as the system is refined) Kin: Kin output is provided in a kinemage format and should be appended to an open kinemage of the structure of interest. This provides validation markup of the structure. Markup takes the form of purple lines, drawn to follow the peptide plane diherdrals used to determine outliers. Clicking on the vertices or on a point in the middle of the center line will bring up validation numbers. The first value corresponds to 'contour_level' in the text output. The values preceded by 'a' and 'rega' correspond to 'loose_alpha' and 'regular_alpha', respectively. The values preceded by 'b' and 'regb' correspond to 'loose_beta' and 'regular_beta', respectively. Points: Points output is provided in kinemage dotlist format. Each point corresponds to one outlier residue and is plotted in the 3D cablam space used to determine outliers (dimensions are CA_pseudodihedral_in, CA_pseudodihedral_out, and peptide_plane_psedudodihedral). These points may be appended to contour kinemages to visualize outliers in that space. This output is intended primarily for developer and exploratory use, rather than for validation per se. (R'R(R)(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pytinterpretationôs =tpdbidcC s8tj|dddddgd|ƒ}tj|ƒ|S(Nt targetatomstCAtCtNtOR=(Rtconstruct_linked_residuesRtcablam_measures(t hierarchyR=tresdata((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_validate.pytsetup:s  cC s dddg}i}x„|D]|}tjjdd|ddtjjƒ}|dkr|tjj d|d ƒtj ƒnt j d |ƒ}|||s^5     4 2 F          A J % $'!=