יxTc@ sddlmZddlZddlZddlmZddlmZddlmZddlm Z ddl m Z ddl Z ddlm Z dd l mZe jjd d eZd Zd ZdZdZdZdZdZejdZeejdZggedejdZejdZggdejdZ dZ!dZ"ejdZ#ejdZ$dZ%dZ&dZ'd Z(d!Z)dS("i(tdivisionN(tpdb(t cablam_res(t cablam_math(tcablam_fingerprints(teasy_run(t file_reader(t group_argss cablam_training { file_or_dir = None .type = path .help = '''input pdb file or dir thereof''' separate_files = False .type = bool .help = '''Generate a separate, auto-named output file for each input file''' give_kin = False .type = bool .help = '''Print output to screen in .kin format (default is comma-separated .csv format)''' give_connections = False .type = bool .help = '''Add prevres and nextres columns to .csv output''' debug = False .type = bool .help = '''Adds some text printed to stderr for debugging esp. for fingerprints''' all_measures = False .type = bool .help = '''Does all measures''' cad = False .type = bool .help = '''2 CA pseudo dihedrals''' caa = False .type = bool .help = '''3 CA pseudo angles''' cod = False .type = bool .help = '''2 CO pseudo dihedrals''' rama = False .type = bool .help = '''2 Ramachandran dihedrals: phi, psi''' exrama = False .type = bool .help = '''4 Ramachandran dihedrals: psi-1, phi, psi, phi+1''' tau = False .type = bool .help = '''1 backbone angle: tau (defined by N-CA-C)''' omega = False .type = bool .help = '''1 backbone dihedral: omega (defined by CA_1-C_1-N_2-CA_2)''' cablam = False .type = bool .help = '''Shortcut for just cablam-relevant measures CA_d_in, CA_d_out, CO_in''' probe_motifs = None .type = strings .help = '''Activates hydrogen bonding analysis, probe=motif_name1,motif_name2,... use --listmotifs to list available fingerprints''' probe_path = None .type = path .help = '''Stores path to dir of probed files, probe will be called for each file if this is not provided''' probe_mode = *kin annote instance sequence superpose .type = choice .help = '''=kin for dotlist kins (default) =annote for ball on model, =instance for vectorlist kins''' list_motifs = False .type = bool .help = '''print motifs/fingerprints available to screen''' b_max = None .type = float .help = '''Set a max b factor, residues containing a backbone atom with higher b will be pruned, recommended: -b=30''' prune_alts = False .type = bool .help = '''Removes all residues with alternate conformations in relevant atoms''' prune = None .type = strings .help = '''List of restypes to be pruned, separated by commas, no spaces eg PRO''' skip_types = None .type = strings .help = '''List of restypes to be skipped during printing, separated by commas''' include_types = None .type = strings .help = '''List of restypes to be printed, all others will be skipped''' cis_or_trans = *both cis trans .type = choice .help = '''selects whether cis-peptides, trans-peptides, or both will be returned''' fear = False .type = bool .help = '''turns on fear-to-tread analysis (this is temporary)''' help = False .type = bool .help = '''print help text to screen''' } tprocess_includescC stjjddS(NsY phenix.cablam_training or cablam_training.py is a program intended for the exploration of protein structure datasets, the annotation of motifs of interest, and the training of reference datasets. It was used in the construction of the reference contours used by cablam_validate. It contains a number of features and modes and is intended primarily as a development tool rather than a utility for typical users. However, anyone interested in exploring protein backboen geometry may find something of use here. -------------------------------------------------------------------------------- file_or_dir=*path* Path to a pdb file or dir of pdb files to operate on, the only argument that doesn't need an explicit flag -------------------------------------------------------------------------------- -----Basic Printing Options----------------------------------------------------- separate_files=True/False Generate a separate, auto-named output file in the current dir for each input file, default output prints a single file to screen give_kin=True/False Print output to screen in .kin format, may be combinded with separate_files, default output prints comma-separated .csv format give_connections=True/False If set to True, adds prevres and nextres columns to .csv output skip_types=restype1,restype2 include_types=restype3,restype4 Together, these control which residue types are printed to screen or file. Default prints all residues. Residue types and relationships given to skip_types are excluded from printing If only include_types is used, only the listed restypes will be printed If include_types and skip_types are both used, then the types given to include_types will override those skipped by skip_types. List restypes by their 3-letter code and separated by commas withoug spaces, e.g. GLY,PRO,ALA,TRP Sequence relationships may be represented with underscores, e.g. _PRO is pre-proline, and GLY__ (2 underscores) is post-post-glycine examples: skip_types=PRO would print every residue except proline include_types=PRO,GLY would print *only* glycines and prolines skip_types=_PRO include_types=GLY would skip pre-prolines unless they were also glycines cis_or_trans='cis' 'trans' 'both' Selects printing for cis-peptides or trans-peptides exclusively. The default is 'both' which will print all residues. cis is defined as -60 to +60 degrees trans is defined as 120 to 180 and -120 to -180 degrees for the omega dihedral Note that selecting 'cis' or 'trans' will also stop printing for any residue for which omega cannot be calculated. -------------------------------------------------------------------------------- -----Probe and Motif Search Options--------------------------------------------- This is an alternate mode which searches for hydrogen bonding patterns defined in fingerprints. probe_motifs=motif_name1,motif_name2 This flag activates hydrogen bonding pattern analysis, which will not run otherwise. The flag accepts a spaceless string of comma-separated motif names to search for. Use list_motifs=True to get a list of available motifs. probe_path=*path* cablam_training can use precomputed probe results to speed up runs on large datasets. If a path to such prepared files is not provided, Reduce and Probe will be run on each pdb file, which may be time-consuming. Running: phenix.probe -u -condense -self -mc -NOVDWOUT -NOCLASHOUT MC filename.pdb > filename.probe Should produce appropriately formatted and named files for this option probe_mode=kin/annote/instance/sequence These are printing options for hydrogen bond pattern analysis, which overrides the Basic Printing Options above. Choose 1 of 3: =kin returns automatically-named kinemage files, one for each unique member residue in each motif. The kins are high-dimensional dotlists containing the measures specified in the commandline (see below for options). This is the default printing. =annote returns an automatically-named kinemage file for each pdb file. These kins are balllists that highlight the selected motifs of interest if appended to existing kinemages of the structures. =instance returns an automatically-named vectorlist kinemage file for each motif of interest. Each kin is a high-dimensional vectorlist that shows the path of a multi-residue motif through the measures specified in the commandline (see below for options) =sequence prints to screen the animo acid sequence of the motif of interest. Does not behave with multiple motifs. Uses single-letter amino acid codes, if a residue type is unrecognized, will print 'X' followed by the 3-letter code. list_motifs=True/False Prints to screen a list of all the motifs/"fingerprints" currently available for hydrogen bond pattern search -------------------------------------------------------------------------------- -----Geometric Measures--------------------------------------------------------- All of these default to False, and some output modes will not function unless at least one of these options is turned on. When in doubt, cablam=True and/or rama=True will provide relevant information. cad=True/False For each residue, calculate the 2 C-alpha pseudo dihedrals caa=True/False For each residue, calculate the 3 C-alpha pseudo angles cod=True/False For each residue, calculate the 2 carbonyl oxygen pseudo dihedrals rama=True/False For each residue, calculate Ramachandran dihedrals phi and psi exrama=True/False For each residue, calculate Ramachandran dihedrals psi-1, phi, psi, phi+1 tau=True/False For each residue, calculate backbone angle tau, defined by N-NA-C omega=True/False For each residue, calculate backbone peptide dihedral, defined by CA_1,C_1,N_2,CA_2 all_measures=True/False For each residue, calculate all of the above measures (may be overkill) cablam=True/False Recommended, but not default behavior. For each residue calculate the measures most relevant to cablam analysis: CA_d_in, CA_d_out, CO_in -------------------------------------------------------------------------------- -----Quality Control Options---------------------------------------------------- b_max=#.# Set a max b factor value. Residues containing a backbone atom with higher b will be pruned and excluded from all calculations. Note this may affect neighboring residues. Strongly Recommenced: b_max=30.0 prune_alts=True/False Prune and excludes from calculations all residues with alternate conformations for backbone atoms. Note this may affect neighboring residues. Default is prune_alts=False, which results in only the first alternate position for each residue being reported on. prune=restype1,restype2 Prune and exclude from calculations the selected list of residue types. Note this may affect neighboring residues. Restypes should be given as 3-letter codes, e.g. GLY,PRO, but this option does not yet support the sequence relationship that skip_types= and include_types= do. -------------------------------------------------------------------------------- -----Help Options--------------------------------------------------------------- help=True/False Displays this help message. list_motifs=True/False Prints to screen a list of all the motifs/"fingerprints" currently available for hydrogen bond pattern search debug=True/False Activates print-to-stderr debugging notes for hydrogen bond pattern search. This may be valuable when trying to define a new pattern correctly and with proper format. -------------------------------------------------------------------------------- Examples: phenix.cablam_training cad=True cod=True skip_types=GLY,PRO,_PRO,ILE,VAL b_max=30.0 kin=True file_or_dir=path/pdbfilename.pdb phenix.cablam_training cablam=True b_max=30.0 prune=GLY probe_motifs=parallel_beta,antiparallel_beta_cwc,antiparallel_beta_wcw probe_mode=kin probe_path=path/database/probefiles file_or_dir=path/database/pdbfiles (tsyststderrtwrite(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pytusages cC s|j}x|D]}t}x}||jD]n}|r=Pnx[||j|D]H}||j|||krO||j|j|}t}PqOqOWq-WqWdS(N(tkeystFalsetaltstatombt removelinkstpoptTrue(tresdatatbmaxtreslisttresidtdeletedtalttatomttrash((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pytstripBEs  cC s\|j}xI|D]A}||}t|jdkr|j|j|}qqWdS(Ni(R tlenRRR(RRRtresidueR((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pyt prune_alts]s     cC s`|rt}n|rt}ntSx|D]}|}|jdrx|jdrxtjjd|dtjnx6|jdr|jr|j}|d}q{tSq{Wx6|jdr|j r|j }|d }qtSqW|j ddk r|j |j dd}ntS||j kr)t}q)q)Wx|D]}|}|jdr|jdrtjjd|dtjnx6|jdr|jr|j}|d}qtSqWx6|jdr|j r|j }|d }qtSqW|j ddk r9|j |j dd}ntS||j krFt}qFqFW|S(Nt_s' Invalid --skip flag argument: s has'_' on both sides iitCAtresname(RRt startswithtendswithR R R texittnextrestprevrestfirstalttNoneRtupper(Rtskiplisttinclisttdoskiptskipt currentresR"tinc((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pyt skipcheckos`                      cC st}|dkrt}nd|jkr3t}ng|jd}|dkrm|dkrm|dkrmt}n|dkr|dks|dkrt}n|S( Ntbothtomegatcisii<ttransixi(RRtmeasures(Rt cis_or_transR-R3((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pytfails_cis_checks    $ $ c C sd}d}d}x't|jD]\}}|}Pq%Wtjj}|j}|j|tjj dt j |d|j } tjj dt j |d| j } dj| j } tjj dt j |d| } | j S( Nsphenix.reduce -quiet -trim -s5phenix.reduce -oh -his -flip -pen9999 -keep -allalt -s?phenix.probe -u -condense -self -mc -NOVDWOUT -NOCLASHOUT ALL -s cleaning . . . t stdin_liness reducing . . . s s probing . . . (t enumeratetmodelsRt hierarchytroott detached_copyt append_modelR R R Rtfully_bufferedt as_pdb_stringt stdout_linestjoin( R<t trim_commandt build_commandt probe_commandtitmtmodeltrtmdct clean_outt build_outt input_strt probe_out((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pytmake_probe_datas"  c C s|j}xv|D]n}|js+qn|jd}|d}|d}|d}|dksjqn|d}|dd!j} t|dd!j} |dd!} |dd !j} | d krqn|d d !} |d d !j}|d}|dd!j}t|dd!j}|dd!}|dd!}|dd !j}|d d !}|d d !j}|d}|d}djd| d| | g}||jkrqn||}djd|d||g}||jkrqn||}|j|}td|d|d|d|d|j|}| |jjkrpi|j| j?| }|j@}tAjB|||}|sqntAjC|d#x|D]}tAjC||qW|jDrtE||jDn|jFrtF|n|j$rI|j%rItGjH|d$t#d%t#nJ|j$rntGjH|d$t#d%tIn%|j%rtGjH|d$tId%t#n|j&rtGjJ|n|j*s|j'rtGjK|n|j(rtGjL|n|j)s|jMd&kr tGjN|n|j+r( tGjO|n|j,r |j.r |jPd'd(}||krtjj=|j.|}tQ|} tR|| | jSq qn|j,r tR|tT|n|j,r t1jU||}!n|j,r |j/d kr tV|!| |n|j/d)krD tQ|d*d+}"tW|!d,|"|"jSn|j/d!kri tX|!|| |nZ|j/d-kr tY|!n>|j/d.kr tZ|!| |}ntj!jd/tjtj}"q|j9rO |j8r3 tQ|d0d+}"t:| |d,|"t[|| |||jMd,|"|"jSq t[|| |||jMq|j8r tQ|d1d+}"t;| |j<d,|"t\|| |||j<|jMd,|"|"jSqt\|| |||j<|jMqW|r x|D]} || jSq WndS(2Nt master_philRRsfile_or_dir="%s"tphiltsourcess s'cctbx_project/mmtbx/cablam/fingerprintss.pickleis-Could not identify valid target file or dir. tCA_d_ins-180 180tCA_d_outtCO_d_intCO_d_outspsi-1tphitpsisphi+1tCA_a_ins0 180tCA_atCA_a_outttauR3R!tORRR~tkinRs not id'd as readable file RSt dodihedralstdoanglesR2s.pdbs.probetannotescablam_motifs.kinRR|tsequencet superposes" Unrecognized probemode request s _cablam.kins _cablam.csv(]tlibtbxRt command_linetargument_interpreterRRRtisfileRtany_filet file_typeRtprocesst file_objectRtfetchtextracttcablam_trainingthelpR R R%t list_motifsRR tlistdirtenvtfind_in_repositoriesR$tsplitextRt file_or_dirR)R t all_measuresRtcadtcaatcodtexramaR R3tramatcablamt probe_motifsRZt probe_patht probe_modeRRtget_all_labelsRtpruneR*t skip_typest include_typestseparate_filestgive_kinRRtgive_connectionsRCRtinputtconstruct_hierarchyRtconstruct_linked_residuest prunerestypetb_maxRRRt CApseudosRtCOpseudodihedralstphipsittaucalcR7t omegacalctcablam_measuresRRR{tcloseRPt check_proteinRRRRRRR(#targst interpreterRRt input_filetarg_philt work_philt work_paramstparamstfilesetRt motifnametdirpathRRt targetatomsRRRt probefilelistt prunelisttresR+R,Rtpdb_intpdb_ioR<Rtrestypet probefilenamet probefilepathR_t found_motifstoutfile((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pytrunsz           "                         "   "                               % # (*t __future__RRR tiotbxRt mmtbx.cablamRRRRRtlibtbx.phil.command_lineRRRtparseRRR RRR1R8RPR{RR}RRRRRRRRRRRRRRV(((sJ/public2/ftp/pub/ccp4/6.5/ccp4-6.5/lib/py2/mmtbx/cablam/cablam_training.pytsD  W   L  % e         ,