o Rŀg5z @sdZddlZddlZddlZddlZddlZddlZddlZddlZddl m Z ddl m Z ddl mZddl mZddl mZddl mZGd d d Zed krd Zeed ZdZdZeejdkrejdZeedZeejdkrejddD]"Zedkrde_q}edkrdZq}edvreddZq}edkrdZq}n eZeZde_eejdkrejddkredeej eeddSejddkrej!ederej"eddSdSejddkrej#eeddSejddkrej"eddSeejddkrFejdd$rFejdZ%ej&e%eeed erBe'Z(e(D]\Z)Z*e)e%kr@ej+e%e*eeed q-dSdSejddd!kre,d"ejdZ-e-D]0Z.ej&e.eeed ere'Z(e(D]\Z)Z*e)e.krej+e.e*eeed qoq[dSdSdSdS)#z?Access the PDB over the internet (e.g. to download structures).N)ThreadPoolExecutor)Optional)Request) urlcleanup)urlopen) urlretrievec@s^eZdZdZdZ    d/ddZedd Zed d Zd d Z ddZ ddZ d0ddZ d1ddZ     d2deededeedeededeef ddZd3ded eeeeffd!d"Z d4d#d$Z   d5d%eedeedeefd&d'Z   d5d%eedeedeefd(d)Z   d5d%eedeedeefd*d+Zd6d-d.ZdS)7PDBListaQuick access to the structure lists on the PDB or its mirrors. This class provides quick access to the structure lists on the PDB server or its mirrors. The structure lists contain four-letter PDB codes, indicating that structures are new, have been modified or are obsolete. The lists are released on a weekly basis. It also provides a function to retrieve PDB files from the server. To use it properly, prepare a directory /pdb or the like, where PDB files are stored. All available file formats (PDB, PDBx/mmCif, PDBML, mmtf) are supported. Please note that large structures (containing >62 chains and/or 99999 ATOM lines) are no longer stored as a single PDB file and by default (when PDB format selected) are not downloaded. Large structures can be downloaded in other formats, including PDBx/mmCif or as a .tar file (a collection of PDB-like formatted files for a given structure). If you want to use this module from inside a proxy, add the proxy variable to your environment, e.g. in Unix: export HTTP_PROXY='http://realproxy.charite.de:888' (This can also be added to ~/.bashrc) a The Protein Data Bank: a computer-based archival file for macromolecular structures. F.C.Bernstein, T.F.Koetzle, G.J.B.Williams, E.F.Meyer Jr, M.D.Brice, J.R.Rodgers, O.Kennard, T.Shimanouchi, M.Tasumi J. Mol. Biol. 112 pp. 535-542 (1977) http://www.pdb.org/. https://files.wwpdb.orgNTcCsPd|_||_|r ||_nt|_||_|r||_n tj|jd|_d|_ dS)zInitialize the class with the default server or a custom one. Argument pdb is the local path to use, defaulting to the current directory at the moment of initialisation. NobsoleteF) assemblies pdb_server local_pdbosgetcwd_verbose obsolete_pdbpathjoin flat_tree)selfserverpdbrverboserC/var/www/html/myenv/lib/python3.10/site-packages/Bio/PDB/PDBList.py__init__Ys   zPDBList.__init__cCs|dur tjddS|S)zPrint a warning to stdout (PRIVATE). Temporary warning (similar to a deprecation warning) that files are being downloaded in mmCIF. NzHWARNING: The default download format has changed from PDB to PDBx/mmCif mmCif)sysstderrwrite file_formatrrr_print_default_format_warningxs z%PDBList._print_default_format_warningcCshtt|#}g}|D]}|}t|dksJ||q Wd|S1s-wY|S)zRetrieve a list of pdb codes in the weekly pdb status file from given URL. Used by get_recent_changes. Typical contents of the list files parsed by this method is now very simply - one PDB name per line. N) contextlibclosingrstriplenappenddecode)urlhandleanswerlinerrrrget_status_lists zPDBList.get_status_listcCs>|jd}||d}||d}||d}|||gS)aReturn three lists of the newest weekly files (added,mod,obsolete). Reads the directories with changed entries from the PDB server and returns a tuple of three URL's to the files of new, modified and obsolete entries from the most recent list. The directory with the largest numerical name is used. Returns None if something goes wrong. Contents of the data/status dir (20031013 would be used);: drwxrwxr-x 2 1002 sysadmin 512 Oct 6 18:28 20031006 drwxrwxr-x 2 1002 sysadmin 512 Oct 14 02:14 20031013 -rw-r--r-- 1 1002 sysadmin 1327 Mar 12 2001 README z/pub/pdb/data/status/latest/z added.pdbz modified.pdbz obsolete.pdb)r r.)rraddedmodifiedr rrrget_recent_changess  zPDBList.get_recent_changescCsf|jd}|jr tdtt|}dd|ddD}Wd|S1s,wY|S)zRetrieve the big file containing all the PDB entries and some annotation. Returns a list of PDB codes in the index file. z'/pub/pdb/derived_data/index/entries.idxz)Retrieving index file. Takes about 27 MB.cSs(g|]}t|dkr|ddqS)r#N)r'r)).0r-rrr s"z+PDBList.get_all_entries..N)r rprintr$r%r readlines)rr*r+ all_entriesrrrget_all_entriess  zPDBList.get_all_entriescCs|jd}tt|+}g}|D]}|dsq|d}t|dks'J||qWd|S1s:wY|S)aReturn a list of all obsolete entries ever in the PDB. Returns a list of all obsolete pdb codes that have ever been in the PDB. Gets and parses the file from the PDB server in the format (the first pdb_code column is the one used). The file looks like this:: LIST OF OBSOLETE COORDINATE ENTRIES AND SUCCESSORS OBSLTE 31-JUL-94 116L 216L ... OBSLTE 29-JAN-96 1HFT 2HFT OBSLTE 21-SEP-06 1HFV 2J5X OBSLTE 21-NOV-03 1HG6 OBSLTE 18-JUL-84 1HHB 2HHB 3HHB OBSLTE 08-NOV-96 1HID 2HID OBSLTE 01-APR-97 1HIU 2HIU OBSLTE 14-JAN-04 1HKE 1UUZ ... z!/pub/pdb/data/status/obsolete.datsOBSLTE r4r#N) r r$r%r startswithsplitr'r(r))rr*r+r r-rrrrget_all_obsoletes      zPDBList.get_all_obsoleteFc Cs||}|}d|d|d|d||dd}||vr1td|dd |d ||}|d vrc|s=d nd }|dkrEdn|dkrKdnd} |jd|d| d|ddd|} n|dkrz|jd|ddd|d|} nd|} |dur|s|jn|j} |jstj | |dd} n|} t | tj st | tj | |} d|d|d|d|d|dd} tj | | |}|stj |r|jrtd|d|S|jrtd |d!z tt| | Wnty td"Y|Swt| d##}t|d$ }||Wdn 1s'wYWdn 1s7wYt| |S)%aFetch PDB structure file from PDB server, and store it locally. The PDB structure's file name is returned as a single string. If obsolete ``==`` True, the file will be saved in a special file tree. NOTE. The default download format has changed from PDB to PDBx/mmCif :param pdb_code: 4-symbols structure Id from PDB (e.g. 3J92). :type pdb_code: string :param file_format: File format. Available options: * "mmCif" (default, PDBx/mmCif file), * "pdb" (format PDB), * "xml" (PDBML/XML format), * "mmtf" (highly compressed), * "bundle" (PDB formatted archive for large structure) :type file_format: string :param overwrite: if set to True, existing structure files will be overwritten. Default: False :type overwrite: bool :param obsolete: Has a meaning only for obsolete structures. If True, download the obsolete structure to 'obsolete' folder, otherwise download won't be performed. This option doesn't work for mmtf format as obsoleted structures aren't stored in mmtf. Also doesn't have meaning when parameter pdir is specified. Note: make sure that you are about to download the really obsolete structure. Trying to download non-obsolete structure into obsolete folder will not work and you face the "structure doesn't exists" error. Default: False :type obsolete: bool :param pdir: put the file in this directory (default: create a PDB-style directory tree) :type pdir: string :return: filename :rtype: string rz.ent.gz.cif.gzz.xml.gzz-pdb-bundle.tar.gz)rrxmlmmtfbundlezSpecified file_format zF does not exist or is not supported. Please use one of the following: z, .)rrr=dividedr rmmCIFXMLz/pub/pdb/data/structures//r?z/pub/pdb/compatible/pdb_bundle/zhttp://mmtf.rcsb.org/v1.0/full/Nz.entz.cifz.xmlz.mmtfz-pdb-bundle.tarStructure exists: '' zDownloading PDB structure ''...zDesired structure doesn't existrbwb)r"lower ValueErrorrr r rrrraccessF_OKmakedirsexistsrr5rrOSErrorgzipopen writelinesremove)rpdb_coder pdirr! overwrite archive_dictarchivepdb_dir file_typer*rfilenamefinal final_filegzoutrrrretrieve_pdb_files .  "        zPDBList.retrieve_pdb_filec Cstj|js Jtj|jrtj|jsJ||}|\}}}||D]B}z#|j||d|rM| }|D]\}} ||krLt j || |ddq;Wq)t yk} zt d|d| dWYd} ~ q)d} ~ ww|D]}|jrtj|jd|d |} |j} ntj|j|d d d|d |} tj|j|d d } tj| d|d |} tj| rtj| dd zt| | Wqnt yt d | dYqnwtj| r|jrt d| dqn|jrt d| dqndS)a,Update your local copy of the PDB files. I guess this is the 'most wanted' function from this module. It gets the weekly lists of new and modified pdb entries and automatically downloads the according PDB files. You can call this module as a weekly cron job. r T)r!rYzerror z:  Nrr@rErF)exist_okzCould not move z to obsolete folderzObsolete file z already movedz is missing)rrisdirr rQrr"r1rcget_all_assembliesplretrieve_assembly_file Exceptionr5rrisfilerPshutilmover)rr!with_assembliesnewr0r rWr a_pdb_code assembly_numerrold_filenew_dirnew_filerrr update_pdb\s^   "    zPDBList.update_pdb pdb_codesr rXr!rYmax_num_threadsc CsV||}t|}|tj|j||||d|WddS1s$wYdS)aJFetch set of PDB structure files from the PDB server and store them locally. :param pdb_codes: A list of 4-symbol PDB structure IDs :param obsolete: Has a meaning only for obsolete structures. If True, download the obsolete structure to 'obsolete' folder. Otherwise, the download won't be performed. This option doesn't work for mmtf format as obsolete structures are not available as mmtf. (default: ``False``) :param pdir: Put the file in this directory. By default, create a PDB-style directory tree. :param file_format: File format. Available options: * "mmCif" (default, PDBx/mmCif file), * "pdb" (format PDB), * "xml" (PMDML/XML format), * "mmtf" (highly compressed), * "bundle" (PDB formatted archive for large structure). :param overwrite: If set to true, existing structure files will be overwritten. (default: ``False``) :param max_num_threads: The maximum number of threads to use when downloading files )r rXr!rYN)r"rmap functoolspartialrc)rrwr rXr!rYrxexecutorrrrdownload_pdb_filess # "zPDBList.download_pdb_filesreturnc Cst|dr|jr|jrtd|jS|jrtdddidd}t|d}d tt|d }t d ||}t |}t | dd }Wd n1sRwYdtdtttffdd}tt||}||_|S)a,Retrieve the list of PDB entries with an associated bio assembly. The requested list will be cached to avoid multiple calls to the server. :param str file_format: A legacy parameter that is left to avoid breaking changes :return: the assemblies :rtype: list r z%Retrieving cached list of assemblies.z7Retrieving list of assemblies. This might take a while.return_all_hitsTassembly)request_options return_typezutf-8zapplication/json; charset=utf-8)z Content-TypezContent-Lengthz+https://search.rcsb.org/rcsbsearch/v2/query result_setNrcSs"|dd}|d|dfS)N identifier-r)r:rL)rr:rrr transformsz-PDBList.get_all_assemblies..transform)hasattrr rr5jsondumpsencodestrr'rrloadsreadr)dicttuplelistry) rr!bodydataheadersrequestresponser rrrrrgs,    zPDBList.get_all_assembliesc Cs|}t|}|d|d|d|dd}||}|}||vr.td|d||}|dkr?|jd |}n|d krL|jd |}ntd |d |durj|j} |jsitj | |dd} n|} t | tj sxt | tj | |} tj | |dd} |stj | r|jrtd| d| S|jrtd|d|dz tt|| Wnty} ztd| WYd} ~ | Sd} ~ wwt| d"} t| d }|| Wdn1swYWdn1swYt| | S)aIFetch one or more assembly structures associated with a PDB entry. Unless noted below, parameters are described in ``retrieve_pdb_file``. :type assembly_num: str :param assembly_num: assembly number to download. :rtype : str :return: file name of the downloaded assembly file. z.pdbz.gzz -assemblyr<)rmmcifzSpecified file_format 'z?' is not supported. Use one of the following: 'mmcif' or 'pdb'.rz#/pub/pdb/data/assemblies/mmCIF/all/rz/pub/pdb/data/biounit/PDB/all/z file_format 'z' not supportedNrErFrGrHzDownloading assembly (z) for PDB entry 'rIz_sz2PDBList.download_all_assemblies..) r"rgrsubmitrzr{rirTrU)rrr!rxr r|outfilerrrdownload_all_assembliesAs$    "zPDBList.download_all_assembliescCs||}|}t|}|tj|j|d|Wdn1s$wY|rMt|d}|dd|DWddS1sFwYdSdS)ahRetrieve all PDB entries not present in the local PDB copy. NOTE: The default download format has changed from PDB to PDBx/mmCif. :param listfile: Filename to which all PDB codes will be written :param file_format: File format. Available options: * "mmCif" (default, PDBx/mmCif file), * "pdb" (format PDB), * "xml" (PMDML/XML format), * "mmtf" (highly compressed), * "bundle" (PDB formatted archive for large structure) :param max_num_threads: The maximum number of threads to use while downloading PDB entries r Nrcs|]}|dVqdSrdNrrrrrrz.PDBList.download_entire_pdb..) r"r8rryrzr{rcrTrUrrr!rxentriesr|rrrrdownload_entire_pdbas   "zPDBList.download_entire_pdbcCs||}|}t|}|tj|jd|d|Wdn1s%wY|rNt|d}|dd|DWddS1sGwYdSdS)a Retrieve all obsolete PDB entries not present in local obsolete PDB copy. NOTE: The default download format has changed from PDB to PDBx/mmCif. :param listfile: Filename to which all PDB codes will be written :param file_format: File format. Available options: * "mmCif" (default, PDBx/mmCif file), * "pdb" (PDB format), * "xml" (PMDML/XML format). :param max_num_threads: The maximum number of threads to use while downloading PDB entries T)r r!Nrcsrrrrrrrrrz4PDBList.download_obsolete_entries..) r"r;rryrzr{rcrTrUrrrrdownload_obsolete_entriess   "z!PDBList.download_obsolete_entriespdb_seqres.txtcCs&|jrtd|jd}t||dS)zKRetrieve and save a (big) file containing all the sequences of PDB entries.z-Retrieving sequence file (takes over 110 MB).z$/pub/pdb/derived_data/pdb_seqres.txtN)rr5r r)rsavefiler*rrrget_seqres_files zPDBList.get_seqres_file)r NNT)FNNF)NF)FNNFN)r~)NNF)NNN)r)__name__ __module__ __qualname____doc__PDB_REFr staticmethodr"r.r1r8r;rcrvrrboolrrr}rrgrirrrrrrrrr6s    % ~B 0+ N " % #r__main__aPDBList.py (c) Kristian Rother 2003, Wiktoria Karwicka & Jacek Smietanski 2016 Contributed to Biopython Usage:: PDBList.py update [options] - write weekly PDB updates to local pdb tree. PDBList.py all [options] - write all PDB entries to local pdb tree. PDBList.py obsol [options] - write all obsolete PDB entries to local pdb tree. PDBList.py assemb [options] - write all assemblies for each PDB entry to local pdb tree. PDBList.py [options] - retrieve single structure PDBList.py (,,...) [options] - retrieve a set of structures Options: -d A single directory will be used as , not a tree. -o Overwrite existing structure files. -pdb Downloads structures in PDB format -xml Downloads structures in PDBML (XML) format -mmtf Downloads structures in mmtf format -with-assemblies Downloads assemblies along with regular entries. Maximum one format can be specified simultaneously (if more selected, only the last will be considered). By default (no format specified) structures are downloaded as PDBx/mmCif files. rFr4)rrFz-dTz-o)z-pdbz-xmlz-mmtfrEz-with-assembliesupdatezupdating local PDB at )r!rnallr obsolassembr#)rXr!rY(z[0-9A-Za-z]{4})/rr$rzrSrrrerlrconcurrent.futuresrtypingrurllib.requestrrrrrrdocr5r!rYrnr'argvpdb_pathrhoptionrrrvrrrisdigitrWrcrgr rprqrifindallpdb_idspdb_idrrrrs#       }    (      9(