:>"^i @s"dZddlZddlmZddlmZddlmZddlmZddlm Z ddl m Z yPdej dd kodknrdd l mZndd l mZWn"ek rdd l mZYnXdZdZejdddZGdddeZdS)zBio.SeqIO support for the "uniprot-xml" file format. See Also: http://www.uniprot.org The UniProt XML format essentially replaces the old plain text file format originally introduced by SwissProt ("swiss" format in Bio.SeqIO). N)Seq) SeqFeature)Alphabet) SeqRecord) as_handle)StringIO) ElementTree) cElementTreez{http://uniprot.org/uniprot}z4%(name)s %(volume)s:%(first)s-%(last)s(%(pub_date)s)Fccst|dw}|jdkr0tdt|tjrRtd|t|tjrt|jtjrtd|t|dst|t rddl }ddl m }|j d |t|}n td tdkrdd l m}|d xitj|d dD]R\}}|dkr,|jtdkr,t|d|d|jV|jq,WWdQRXdS)aIterate over UniProt XML as SeqRecord objects. parses an XML entry at a time from any UniProt XML file returns a SeqRecord for each iteration This generator can be used in Bio.SeqIO return_raw_comments = True --> comment fields are returned as complete XML to allow further processing skip_parsing_errors = True --> if parsing errors are found, skip to next entry rUz Empty file.zWrong alphabet %rreadrN)BiopythonDeprecationWarningz/Passing an XML-containing handle is recommendedzMRequires an XML-containing handle (or XML as a string, but that's deprecated))MissingExternalDependencyErrorzmNo ElementTree module was found. Use Python 2.5+, lxml or elementtree if you want to use Bio.SeqIO.UniprotIO.eventsstartendentryalphabetreturn_raw_comments)zstartzend)rreadline ValueError isinstancerZNucleotideAlphabetZGappedrhasattrstrwarningsBiorwarnr TypeErrorr r iterparsetagNSParserparseclear)handlerrrrreventelemr*8/tmp/pip-build-ww9dw3qa/biopython/Bio/SeqIO/UniprotIO.pyUniprotIterator,s6      "r,c@s7eZdZdZejdddZddZdS)r$zParse a UniProt XML entry to a SeqRecord. return_raw_comments=True to get back the complete comment field in XML format alphabet=Alphabet.ProteinAlphabet() can be modified if needed, default is protein alphabet. FcCs||_||_||_dS)zInitialize the class.N)rrr)selfr)rrr*r*r+__init__ms  zParser.__init__cs'jjtdkstfddfdd}fdd}fdd }fd d }fd d }fdd}fdd}fdd}fdd} fdd} fdd} dddfdd} fdd } fd!d"}fd#d$}td%d&d%_jjjd'd(_xUjjj D]A\}}|d)krt |jj |.append_to_annotationscs4|jj_jjjjd|jdS)N:)textr/namedbxrefsr1dbname)element)r-r*r+ _parse_namesz!Parser.parse.._parse_namecs5d|jjjjjd|jdS)N accessionsr5)r6r/r8r1r9)r:)r4r-r*r+_parse_accessions z&Parser.parse.._parse_accessioncsd}x|D]}|jtdtdtdgkrx|D]s}d|jjtd|jjtdf}||j|jtdkr>| r>|jj_d}q>Wq |jtd krq |jtd kr q Wd S) zParse protein names (PRIVATE).FZrecommendedNameZ submittedNameZalternativeNamez%s_%srZfullNameT componentdomainN)r"r#replacer6r/ description)r:Z descr_setZprotein_elementZrec_nameann_key)r4r-r*r+_parse_proteins$   z$Parser.parse.._parse_proteincsx||D]t}d|jkrd|jjtd|jdf}|jddkrk|jjj|._parse_genecsd|jddS)N geneLocationrD)rE)r:)r4r*r+_parse_geneLocationsz)Parser.parse.._parse_geneLocationcs`d}}}x|D]}|jtdkr|jr|jddkrV|j}q|jddkru|j}qd|jq|jtdkrjjj|jdd|jd q|jtd krx1|D])}|jtd krd |jqWqW|r1|r1d ||f}n|r@|}n |rL|}|jjd._parse_organismcs8x1|D])}|jtdkrd|jqWdS)Nr7Z organism_host)r"r#r6)r:rN)r4r*r+_parse_organismHosts z)Parser.parse.._parse_organismHostcsd|jdS)Nkeywords)r6)r:)r4r*r+_parse_keywordsz$Parser.parse.._parse_keywordcsddddddddd d d d d ddddddddddg}|jd|krd|jdjdd}x|jtdD]}|jr||jqWn|jddkrEx|jtdD]b}xY|D]Q}|jrd|jdjdd|jjtdf}||jqWqWn|jdd krx|jtd!D]+}d"|jd}||jd#qlWn|jdd$krx|jtd%D]Q}d&|jdjdd}x+|jtd'D]}||jqWqWn/|jdd(krd|jdjdd}d)} } x|jtd*D]}t|jtd+} y| rt| d)jd+} | d,} n^tt|jtd-d)jd+} | d,8} tt|jtd.d)jd+} Wqktt fk r0YqkXqkW|jd/} |jd0} | | kofd)knr|d1| | fqL|d2| | | | fn|jdd3krn|jdd4krLx{|jtd5D]f}d|jdjdd}x@|jtd5D]+}|d6|jd7|jd8fqWqWj rd9|jdjdd}|t j |d:S);acParse comments (PRIVATE). Comment fields are very heterogeneus. each type has his own (frequently mutated) schema. To store all the contained data, more complex data structures are needed, such as annotated dictionaries. This is left to end user, by optionally setting: return_raw_comments=True The original XML is returned in the annotation fields. Available comment types at december 2009: - "allergen" - "alternative products" - "biotechnology" - "biophysicochemical properties" - "catalytic activity" - "caution" - "cofactor" - "developmental stage" - "disease" - "domain" - "disruption phenotype" - "enzyme regulation" - "function" - "induction" - "miscellaneous" - "pathway" - "pharmaceutical" - "polymorphism" - "PTM" - "RNA editing" - "similarity" - "subcellular location" - "sequence caution" - "subunit" - "tissue specificity" - "toxic dose" - "online information" - "mass spectrometry" - "interaction" ZallergenZ biotechnologyzbiophysicochemical propertieszcatalytic activityZcautionZcofactorzdevelopmental stageZdiseaser?zdisruption phenotypezenzyme regulationfunctionZ inductionZ miscellaneousZpathwayZpharmaceuticalZ polymorphismZPTMz RNA editingZ similarityZsubunitztissue specificityz toxic doserDz comment_%s rr6zsubcellular locationZsubcellularLocationz comment_%s_%sZ interactionZ interactantzcomment_%s_intactIdZintactIdzalternative productsZisoformzcomment_%s_isoformrLzmass spectrometryrlocationpositionr beginrmassmethodzundefined:%s|%sz %s..%s:%s|%szsequence cautionzonline informationlinkz%s@%sr7urizcomment_%s_xmlN) rEr@ getiteratorr#r6r"listintrKeyErrorrr tostring)r:Zsimple_commentsrBZ text_elementZsubloc_elementelZinteract_elementZ alt_elementZ id_elementrrZpos_elsrXrYZ link_element)r4r-r*r+_parse_comments,       ' +   $ z$Parser.parse.._parse_commentc sjjj|jdd|jdd|jkr|jddkrd}d}xo|D]g}|jtdkr^|jd}|dkr|jd}|d kr|jd}|d kr^|jdjd }x|D]}|jjd }|d dkrtj}|jd|_ |jd|j d<||j d<||j d <|djd|j d ._parse_dbReferencec s8tj}g}g}g}d}d}d}x"|D]}|jtdkr|jd}|dkr~|d|jd7}d|jkr|jd}|jjdd}|jjd d} |jjd d} |jjd d} x`|D]} | jtd kr| j|_q| jtd krXx| D]} |j| jdq7Wq| jtdkrj j j| jdd| jd| jddkr| jd|_ q|jddkr| jd|_ qWq7|jtdkr |j|jq7|jtdkr7x1|D])}|jtdkr$|j|jq$Wq7W|rqddj |}nd}|rddj |}nd}g|_dj ||_|r |r| r| r| rtd|d | d | d | d|i|_n ||_dj ||||f|_d|dS)NrZcitationrDZ submissionz to the dbr7datevolumefirstlasttitleZ authorListrKr5rLZPubMedZMEDLINEZscopesourceZtissuezScope: z, zTissue: pub_datez | Z references)rZ Referencer"r#rEgetr6rwr1r/r8Z pubmed_idZ medline_idjoinrUauthorsREFERENCE_JOURNALZjournalcomment)r: referencer|ZscopesZtissuesZ journal_nameZpub_typeryrmZj_volumeZj_firstZj_lastZ cit_elementZperson_elementZsource_elementZ scopes_strZ tissues_str)r4r-r*r+_parse_referencesr           z&Parser.parse.._parse_referencercSsyt|jd|}Wn+tk rH}z d}WYdd}~XnX|jjdd}|dkr|dks|ttjS|stj|S|dkrtj|S|dkrtj |S|dkrtj |St d|dS) NrVstatusrunknownz greater thanz less thanZ uncertainzPosition status %r) r^rEr_rzAssertionErrorrZUnknownPositionZ ExactPositionZ AfterPositionZBeforePositionZUncertainPositionNotImplementedError)r:offsetrVerrrr*r*r+_parse_positions"         z%Parser.parse.._parse_positionc stj}x*|jjD]\}}||j|._parse_featurecsd|jddS)NproteinExistencerD)rE)r:)r4r*r+_parse_proteinExistencesz-Parser.parse.._parse_proteinExistencecs7x0|jjD]\}}|}||qWdS)N)rEr)r:rrrB)r4r*r+_parse_evidencesz%Parser.parse.._parse_evidencecsxZ|jjD]I\}}|dkrEt|jjd|._parse_sequencerrLZdatasetZUnknownDatasetrr7Z accessionZproteinZgenerGrMZ organismHostkeywordr~rKrrprZevidencesequencer<)rr"r#rrr/rErzr9rr^r0sortedsetr8rL)r-r;r=rCrFrHrOrPrRrbrqrrrrrrrr:r*)rr4r-r+r%usx 0G                  z Parser.parseN)__name__ __module__ __qualname____doc__rgeneric_proteinr.r%r*r*r*r+r$fs r$)rr)rr )rsysrrrrZ Bio.SeqRecordrZBio.FilerZ Bio._py3kr version_infoZ xml.etreer r ImportErrorr#r}rr,objectr$r*r*r*r+s" ) 9