B ‰°bŽ<ã@sðdZddlZy ddlZWn(ek r@ddlmZedƒ‚YnXddlmZddlm Z ddl m Z ddl m Z dd lmZmZmZmZdd lmZmZdd lmZdeeed œdd„Zdd„Zdd„Zdd„Zeedœdd„Zddd„ZdS)zDPICIO: read and write Protein Internal Coordinate (.pic) data files.éN)ÚMissingPythonDependencyErrorz:Install NumPy to build proteins from internal coordinates.)Ú as_handle)ÚStructureBuilder)Ú_parse_pdb_header_list)Ú PDBException)Ú IC_ResidueÚIC_ChainÚEdronÚAtomKey)ÚDictÚTextIO)Ú StructureF)ÚfileÚverboseÚreturncCsàt d¡}t d¡}t d¡}t d¡}t d¡}t d¡}tƒ}tgƒ} ddddg} |j|j|j|jg} d} t|dd  } x|   ¡D]}|  d ¡rœqŠ|  d ¡rô|  |¡}|rà|  d ¡| d <|  d¡| d<|  d¡| d<n|ròt d|d|ƒqŠ|  d¡r:|  |¡}|r$|  d¡ ¡| d<n|r”t d|d|ƒqŠ|  d¡r–|  |¡}|r||  d¡}|dkrnd}|  d¡t|  d¡ƒ|  d¡|g}| |kr xptdƒD]d}| |||kr¤| |||ƒ||| |<d|krì| | ¡nd|kr¤d| d<| d<q¤W| |  d ¡|  d!¡t|  d"¡ƒ|  d#¡¡|j} d|  ¡krpx"| j ¡D]}|jsX|} PqXWt| ƒ| _n|rt d|d$|ƒdSqŠ|  d%¡r®|  |¡}|r”| dkrÔ|rÐt d|d&|ƒdS| j|  d ¡ks| jdt|  d"¡ƒkr&|r"t d|d'| jt| jƒd(|ƒdSt t|  d)¡ƒt|  d*¡ƒt|  d+¡ƒfd,¡}| |  d-¡ ¡|t|  d.¡ƒt|  d/¡ƒ|  d0¡|  d-¡t|  d1¡ƒ|  d2¡ ¡¡qŠ|  d3¡r,|  |¡}|r”xÈ|  ¡D]R}|dk rÔ|  |¡}|rÔ| dk rÔ| jrÔ| j}t|  d¡ƒ|j!|  d¡<qÔWqŠt"j#  |¡}|rZ| dk rZ| j $| %¡¡qŠ|rrt d4|d5|ƒdS| ¡rŠ|rt d6|d7|d8ƒdSqŠWWdQRX| &¡}x0| '¡D]$}t(|ƒ}|_| )¡| *¡q´W|S)9a¡Load Protein Internal Coordinate (.pic) data from file. PIC file format: - comment lines start with # - (optional) PDB HEADER record - idcode and deposition date recommended but optional - deposition date in PDB format or as changed by Biopython - (optional) PDB TITLE record - repeat: - Biopython Residue Full ID - sets residue IDs of returned structure - (optional) PDB N, CA, C ATOM records for chain start - (optional) PIC Hedra records for residue - (optional) PIC Dihedra records for residue - (optional) BFAC records listing AtomKeys and b-factors An improvement would define relative positions for HOH (water) entries. N.B. dihedron (i-1)C-N-CA-CB is ignored in assembly if O exists. C-beta is by default placed using O-C-CA-CB, but O is missing in some PDB file residues, which means the sidechain cannot be placed. The alternate CB path (i-1)C-N-CA-CB is provided to circumvent this, but if this is needed then it must be adjusted in conjunction with PHI ((i-1)C-N-CA-C) as they overlap. (i-1)C-N-CA-CB is included by default in .pic files for consistency and informational (e.g. statistics gathering) purposes, as otherwise the dihedron would only appear in the few cases it is needed for. :param Bio.File file: file name or handle :param bool verbose: complain when lines not as expected :returns: Biopython Structure object, Residues with .internal_coord attributes but no coordinates except for chain start N, CA, C atoms if supplied, **OR** None on parse fail (silent unless verbose=True) zm^HEADER\s{4}(?P.{1,40})(?:\s+(?P
\d\d\d\d-\d\d-\d\d|\d\d-\w\w\w-\d\d))?(?:\s+(?P[0-9A-Z]{4}))?\s*$z^TITLE\s{5}(?P.+)\s*$z³^\('(?P[^\s]*)',\s(?P\d+),\s'(?P\s|\w)',\s\('(?P\s|[\w\s-]+)',\s(?P-?\d+),\s'(?P\s|\w)'\)\)\s+(?P[\w]{1,3})(\s\[(?P[a-zA-z\s]+)\])?\s*$a*^ATOM\s\s(?:\s*(?P\d+))\s(?P[\w\s]{4})(?P\w|\s)(?P[\w]{3})\s(?P.)(?P[\s\-\d]{4})(?P[A-Za-z\s])\s\s\s(?P[\s\-\d\.]{8})(?P[\s\-\d\.]{8})(?P[\s\-\d\.]{8})(?P[\s\d\.]{6})(?P[\s\d\.]{6})\s{6}(?P[a-zA-z\s]{4})(?P.{2})(?P.{2})?\s*$z|^BFAC:\s([^\s]+\s+[\-\d\.]+)\s*([^\s]+\s+[\-\d\.]+)?\s*([^\s]+\s+[\-\d\.]+)?\s*([^\s]+\s+[\-\d\.]+)?\s*([^\s]+\s+[\-\d\.]+)?z([^\s]+)\s+([\-\d\.]+)NÚr)Úmodeú#zHEADER ZcfÚheadÚidÚidcodeÚddÚdeposition_datezReading pic filezHEADER parse fail: zTITLE ZttlÚnamezTITLE parse fail:, ú(é z éééérÚresZhetÚposZicodezresidue ID parse fail: zATOM z"ATOM without residue configured:, z ATOM not in configured residue (z):ÚxÚyÚzÚfÚatmZtfacZoccZalcZserZelmzBFAC: z PIC file: z9 error: no residue info before reading (di/h)edron data: zReading PIC filezparse fail on: .Ú.)+ÚreÚcompilerrZinit_structureZ init_modelZ init_chainZinit_segrÚ readlinesÚ startswithÚmatchÚgroupÚprintÚstripÚintÚrangeZ set_headerZ init_residueZresidueÚ is_disorderedÚ child_dictÚvaluesÚinternal_coordrZresnamerÚstrÚnumpyZarrayÚfloatZ init_atomÚgroupsZbfactorsr Zedron_reZload_PICÚ groupdictZ get_structureZ get_chainsrZ link_residuesZ init_edra)rrZ pdb_hdr_reZ pdb_ttl_reZ biop_id_reZ pdb_atm_reZbfac_reZbfac2_reZstruct_builderZ header_dictZ curr_SMCSZ SMCS_initZsb_resZhandleZalineÚmZsegidZ this_SMCSÚirZcoordZ bfac_pairZm2ZrpÚstructÚchnZchnp©r?ú,lib/python3.7/site-packages/Bio/PDB/PICIO.pyÚread_PICs$             "            &        " rAcCs`|jrL|r|s6|j}|s|j}|s6|jj}|j d¡}| |j ||¡¡n| t |¡¡dS)Nr) r5ÚparentrÚheaderÚgetÚwriteÚ write_PICrZ_residue_string)ÚentityÚfpÚpdbidÚchainidÚchainr=r?r?r@Ú_wprs rLcCsªd}x| ¡D]}| ¡ }PqW|r¦d}xz| ¡D]n}d| ¡kr~x\|j ¡D](}x"| ¡D]}| |¡|d7}q^WqPWq4x"| ¡D]}| |¡|d7}qˆWq4WdS)NFrr)Z get_atomsZget_serial_numberZ get_residuesr2r3r4Zget_unpacked_listZset_serial_number)rGZneedr&Zanumr rr?r?r@Ú_enumerate_entity_atomss    rMcCsDx| ¡r| ¡}qWd|jkr8x|D] }t|ƒq&Wnt|ƒdS)z2Ensure all atoms in entity have serial_number set.ÚSN)Z get_parentÚlevelrM)rGÚmdlr?r?r@Úenumerate_atoms)s     rQ)Údatestrrc Csl|rht d|¡}|rhdddddddd d d d d dg t| d¡ƒ}| d¡d|d| d¡dd…}|S)z'Convert yyyy-mm-dd date to dd-month-yy.z(\d{4})-(\d{2})-(\d{2})ZXXXZJANZFEBZMARZAPRZMAYZJUNZJULZAUGZSEPZOCTZNOVZDECrrú-réþÿÿÿN)r(r,r0r-)rRr;Zmor?r?r@Úpdb_date5s& (rUc CsÒt|ƒt|dƒ´}y„d|jkr0tdƒ‚njd|jkr|d| ¡krjx0|j ¡D]}t||||ƒqRWnt||||ƒnd|jkr®|s|j}x|D]}t ||||ƒq–Wnìd|jkrÖxà|D]}t ||||ƒq¾WnÄd|jkrˆ|sô|j   d d ¡}|j   d d ¡}t |j   d d ¡ƒ} |r>|  d  | ¡| p0d|p8d¡¡|j   dd ¡} | rh|  d|  ¡d¡x0|D]} t | |||ƒqnWntdt|jƒƒ‚Wn&tk rÂtdt|ƒƒ‚YnXWd QRXd S)aWrite Protein Internal Coordinates (PIC) to file. See read_PIC() for file format. Recurses to lower entity levels (M, C, R). :param Entity entity: Biopython PDB Entity object: S, M, C or R :param Bio.File file: file name or handle :param str pdbid: PDB idcode, read from entity if not supplied :param char chainid: PDB Chain ID, set from C level entity.id if needed :raises PDBException: if entity level not S, M, C, or R :raises Exception: if entity does not have .level attribute ÚwÚAzNo PIC output at Atom levelÚRrÚCÚMrNrNrrzHEADER {:40}{:8} {:4} Úrz TITLE Ú zCannot identify level: z2write_PIC: argument is not a Biopython PDB Entity )rQrrOrr2r3r4rLrrFrCrDrUrEÚformatÚupperr6ÚKeyErrorÚ Exception) rGrrIrJrHrr r>ZhdrrZnamrPr?r?r@rFMsH           rF)F)NN) Ú__doc__r(r7Ú ImportErrorZBiorZBio.FilerZBio.PDB.StructureBuilderrZBio.PDB.parse_pdb_headerrZBio.PDB.PDBExceptionsrZBio.PDB.internal_coordsrrr r Útypingr r ZBio.PDB.Structurer ÚboolrArLrMrQr6rUrFr?r?r?r@Ús*        i