XfhdZdZdZGddeZdZedkrddlmZed Sd S) aXParser for the cellosaurus.txt file from ExPASy. See https://web.expasy.org/cellosaurus/ Tested with the release of Version 18 (July 2016). Functions: - read Reads a file containing one cell line entry - parse Reads a file containing multiple cell line entries Classes: - Record Holds cell line data. Examples -------- This example downloads the Cellosaurus database and parses it. Note that urlopen returns a stream of bytes, while the parser expects a stream of plain string, so we use TextIOWrapper to convert bytes to string using the UTF-8 encoding. This is not needed if you download the cellosaurus.txt file in advance and open it (see the comment below). >>> from urllib.request import urlopen >>> from io import TextIOWrapper >>> from Bio.ExPASy import cellosaurus >>> url = "ftp://ftp.expasy.org/databases/cellosaurus/cellosaurus.txt" >>> bytestream = urlopen(url) >>> textstream = TextIOWrapper(bytestream, "UTF-8") >>> # alternatively, use >>> # textstream = open("cellosaurus.txt") >>> # if you downloaded the cellosaurus.txt file in advance. >>> records = cellosaurus.parse(textstream) >>> for record in records: ... if 'Homo sapiens' in record['OX'][0]: ... print(record['ID']) # doctest:+ELLIPSIS ... #15310-LN #W7079 (L)PC6 0.5alpha ... c#8K t|}|sdS|V)zParse cell line records. This function is for parsing cell line files containing multiple records. Arguments: - handle - handle to the file. TN)__read)handlerecords 6lib/python3.11/site-packages/Bio/ExPASy/cellosaurus.pyparser2s1  E cnt|}|}|rtd|S)zRead one cell line record. This function is for parsing cell line files containing exactly one record. Arguments: - handle - handle to the file. z$More than one cell line record found)rread ValueError)rr remainders rr r Cs9F^^F IA?@@@ Mrc$eZdZdZdZdZdZdS)Recorda$Holds information from an ExPASy Cellosaurus record as a Python dictionary. Each record contains the following keys: ========= ============================== ======================= Line code Content Occurrence in an entry ========= ============================== ======================= ID Identifier (cell line name) Once; starts an entry AC Accession (CVCL_xxxx) Once AS Secondary accession number(s) Optional; once SY Synonyms Optional; once DR Cross-references Optional; once or more RX References identifiers Optional: once or more WW Web pages Optional; once or more CC Comments Optional; once or more ST STR profile data Optional; twice or more DI Diseases Optional; once or more OX Species of origin Once or more HI Hierarchy Optional; once or more OI Originate from same individual Optional; once or more SX Sex of cell Optional; once AG Age of donor at sampling Optional; once CA Category Once DT Date (entry history) Once // Terminator Once; ends an entry ========= ============================== ======================= ct|d|d<d|d<d|d<d|d<g|d<g|d<g|d<g|d <g|d <g|d <g|d <g|d <g|d<d|d<d|d<d|d<d|d<dS)zInitialize the class.IDACASSYDRRXWWCCSTDIOXHIOISXAGCADTN)dict__init__selfs rr#zRecord.__init__ss dT T T T T T T T T T T T T T T T T rc|drA|dr!|jjd|dd|ddS|jjd|ddS|jjdS)z@Return the canonical string representation of the Record object.rrz (z, )z ( )) __class____name__r$s r__repr__zRecord.__repr__s{ : 4Dz C.1PPT$ZPP4:PPPP.1BBT$ZBBBBn-333 3rcd|dz}|d|dzz }|d|dzz }|d|dzz }|d t|d zz }|d t|d zz }|d t|dzz }|dt|dzz }|dt|dzz }|dt|dzz }|dt|dzz }|dt|dzz }|dt|dzz }|d|dzz }|d|dzz }|d|d zz }|d!|d"zz }|S)#z=Return a readable string representation of the Record object.zID: rz AC: rz AS: rz SY: rz DR: rz RX: rz WW: rz CC: rz ST: rz DI: rz OX: rz HI: rz OI: rz SX: rz AG: rz CA: r z DT: r!)repr)r%outputs r__str__zRecord.__str__s$t*$'DJ&&'DJ&&'DJ&&'Dd,,,,'Dd,,,,'Dd,,,,'Dd,,,,'Dd,,,,'Dd,,,,'Dd,,,,'Dd,,,,'Dd,,,,'DJ&&'DJ&&'DJ&&'DJ&& rN)r) __module__ __qualname____doc__r#r*r.rrrrUsK:*444rrcd}|D]}|dd|dd}}|dkrt}||d<B|dvr||xx|z cc<W|dvr|||w|dkrZ|d\}}|d||f|dkr|r|cS|rt d dS) Nr)rrrrrr r!)rrrrrrrrr;z//zUnexpected end of stream)rstriprappendsplitstripr )rrlinekeyvaluekvs rrrs< F"1"XtABBx00U $;;XXF F4LL > > > 3KKK5 KKKK      3K  u % % % % D[[;;s##DAq 4L  AGGII 6 7 7 7 7 D[[    5344455r__main__) run_doctestN) r1rr r"rrr) Bio._utilsrBr2rrrDs ))X"$PPPPPTPPPl555D z&&&&&&KMMMMMr