o Rŀgd@sddZddlmZddlmZddlmZddlmZddl m Z dZ dZ dd d Z Gd ddZd S)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). ) ElementTree)errors) SeqFeature)Seq) SeqRecordz{http://uniprot.org/uniprot}z4%(name)s %(volume)s:%(first)s-%(last)s(%(pub_date)s)NFc cs|dur tdzItj|ddD]>\}}|dkr7|dds7td|dd ks7td td |dd |d krO|jtd krOt||dV|qWdStj yu}zt j |j t j krp|jdkskJtddd}~ww)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 Argument source is a file-like object or a path to a file. Optional argument alphabet should not be used anymore. 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 N,The alphabet argument is no longer supported)startstart-nsend)eventsr zhttp://www.w3.org/{}z;SeqIO format 'uniprot-xml' only parses xml with namespace: z but xml has namespace: {r entry)return_raw_comments)r rz Empty file.) ValueErrorr iterparse startswithNStagParserparseclear ParseErrorrmessagescodeXML_ERROR_NO_ELEMENTSposition)sourcealphabetreventelem exceptionr#G/var/www/html/myenv/lib/python3.10/site-packages/Bio/SeqIO/UniprotIO.pyUniprotIterators4   r%c@s"eZdZdZdddZddZdS) rzParse a UniProt XML entry to a SeqRecord. Optional argument alphabet is no longer used. return_raw_comments=True to get back the complete comment field in XML format NFcCs |durtd||_||_dS)zInitialize the class.Nr)rrr)selfr!rrr#r#r$__init__Ls zParser.__init__csjjtdks Jfddfdd}fdd}fdd }fd d }fd d }fdd}fdd}fdd}fdd} fdd} fdd} d;ddfdd} fdd } fd!d"}fd#d$}td%d&d'_jjd(d)_jjD]\}}|d*vrt |jj |<q|jj |<qjD]}|jtd+kr||q|jtd,kr||q|jtd-kr||q|jtd.kr||q|jtd/kr||q|jtd0kr||q|jtd1kr||q|jtd2kr ||q|jtd3kr| |q|jtd4kr$| |q|jtd5kr1| |q|jtd6kr>| |q|jtd7krK| |q|jtd8krX||q|jtd9kre||q qt t jj j_ jjsjj d:dj_jS).append_to_annotationscs(|jj_jjjd|jdS)N:)textr)namedbxrefsr+dbnameelementr.r#r$ _parse_name]s z!Parser.parse.._parse_namecs*d|jjjjd|jdS)N accessionsr0)r1r)r3r+r4r5r/r&r#r$_parse_accessionasz&Parser.parse.._parse_accessioncsd}|D]L}|jtdtdtdfvr@|D](}d|jtd|jtdf}||j|jtdkr>|s>|jj_d}qq|jtd krHq|jtd krP qd S) zParse protein names (PRIVATE).FrecommendedName submittedNamealternativeNamez%s_%sfullNameT componentdomainN)rrreplacer1r) description)r6 descr_setprotein_elementrec_nameann_keyr9r#r$_parse_proteings0     z$Parser.parse.._parse_proteincs^|D]*}d|jvr,d|jtd|jdf}|jddkr&|jjj|<q||jqdS)Ntypez gene_%s_%sr>primary)attribrrBrr1r)r*)r6genename_elementrGr9r#r$ _parse_genes   z!Parser.parse.._parse_genecd|jddS)N geneLocationrIrKr5r/r#r$_parse_geneLocationz)Parser.parse.._parse_geneLocationcsd}}}|D][}|jtdkr1|jr0|jddkr|j}q|jddkr*|j}qd|jq|jtdkrJjj|jdd|jd q|jtd krc|D]}|jtd krbd |jqSq|rq|rq|d |d}n |rv|}n|rz|}|jjd<dS)Nr>r2rI scientificcommon organism_name dbReferencer0idlineagetaxontaxonomyz ()organism)rrr1rKr)r3r+r*)r6rVcom_namesci_nameorganism_element taxon_elementr9r#r$_parse_organisms@  z%Parser.parse.._parse_organismcs(|D]}|jtdkrd|jqdS)Nr2 organism_host)rrr1)r6r`rQr#r$_parse_organismHosts  z)Parser.parse.._parse_organismHostcsd|jdS)Nkeywords)r1r5rQr#r$_parse_keywordsz$Parser.parse.._parse_keywordc s\gd}|jd|vr,d|jddd}|tdD] }|jr)||jqne|jddkr_|tdD]"}|D]}|jr[d |jddd|jtdf}||jq>q:n2|jdd kr|td D]}d|jdd }||jd qmn|jddkr|tdD]}d|jddd}|tdD]}||jqqn|jddkrNd|jddd}d} } |tdD]I}t|td} z2| rt| djd} | d} n tt|tdjd} | d8} tt|tdjd} Wqt t fyYqw|jd} |jd} | | kr/dkr=nn |d| d| nT|| d| d| d| nC|jddkrWn:|jdd kr|td!D]*}d|jddd}|td!D]}||jd"d#|jd$q{qfj rd|jdddd%}|t |d&Sd&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" )allergen biotechnologyzbiophysicochemical propertieszcatalytic activitycautioncofactorzdevelopmental stagediseaserAzdisruption phenotypezenzyme regulationfunction induction miscellaneouspathwaypharmaceutical polymorphismPTMz RNA editing similaritysubunitztissue specificityz toxic doserIcomment_ r>r1zsubcellular locationsubcellularLocationz comment_%s_%s interaction interactant _intactIdintactIdzalternative productsisoformzcomment_%s_isoformrXzmass spectrometryrlocationrr beginr massmethodz undefined:|z..r0zsequence cautionzonline informationlinkr2@uri_xmlN)rKrBiterrr1rlistintnextrKeyErrorrrtostring)r6simple_commentsrG text_elementsubloc_elementelinteract_element alt_element id_elementrr pos_elsrr link_elementr9r#r$_parse_comments+       "z$Parser.parse.._parse_commentc sljj|jdd|jdd|jvr|jddkrd}d}|D]}|jtdkr|jd}|dkr:|jd}|d krC|jd}|d kr|jdd }|D]T}|d }|d dkrt}|jd|_ |jd|j d<||j d<||j d <|dd|j d <t |d ddd } t |d dd } t | | |_ qQq#|D] }|jtdkr qdS)NrIr0rXPDBr>propertyrr- resolutionchains,=r -r2r/)r)r3r+rKrrsplitstriprrI qualifiersrSimpleLocationr}) r6rr ref_elementdat_typepairsr!pairfeaturerr r.r#r$_parse_dbReference=sP         z(Parser.parse.._parse_dbReferencecsJt}g}g}g}d}d}d}|D]}|jtdkr|jd}|dkr-|d|jd7}d|jvr7|jd}|jdd}|jd d} |jd d} |jd d} |D]V} | jtd krc| j|_qU| jtd krx| D] } || jdqlqU| jtdkrj j | jdd| jd| jddkr| jd|_ qU|jddkr| jd|_ qUq|jtdkr||jq|jtdkr|D]}|jtdkr||jqq|rdd |}nd}|rdd |}nd}g|_d ||_|r|r| r| r| rt|| | | |d|_n||_d ||||f|_d|dS)Nr>citationrI submissionz to the dbr2datevolumefirstlasttitle authorListrWr0rXPubMedMEDLINEscopertissuezScope: z, zTissue: )r2rrrpub_datez | references)r ReferencerrrKgetr1rr+r)r3 pubmed_id medline_idjoinr}authorsREFERENCE_JOURNALjournalcomment)r6 referencerscopestissues journal_namepub_typerrj_volumej_firstj_last cit_elementperson_elementsource_element scopes_str tissues_strr9r#r$_parse_referencems         z&Parser.parse.._parse_referencercSsz t|jd|}Wn tyd}Ynw|jdd}|dkr,|dus(JtS|s3t|S|dkrunknownz greater thanz less than uncertainzPosition status ) rrKrrrUnknownPosition ExactPosition AfterPositionBeforePositionUncertainPositionNotImplementedError)r6offsetrrr#r#r$_parse_positions$      z%Parser.parse.._parse_positionc st}|jD] \}}||j|<q |jdd|_d|jvr&|jd|_|D]Y}|jtdkrk| td}|rH|d}|d}|}n| tdd}|d}| td d}|}t |||_ q(z |j |j|j td<Wq(tyYq(wjj|dS) NrIr>rXr}rrr~r )rrKitemsrrrIrXrrfindallrr}r1rB Exceptionr)featuresr+)r6rkvfeature_elementposition_elementsstart_position end_position)rr&r#r$_parse_features6       z$Parser.parse.._parse_featurecrN)NproteinExistencerIrPr5rQr#r$_parse_proteinExistencerSz-Parser.parse.._parse_proteinExistencecs&|jD] \}}|}||qdSr()rKr)r6rrrGrQr#r$_parse_evidences z%Parser.parse.._parse_evidencecsn|jD]\}}|dvrt|jjd|<q|jjd|<qtd|jj_ djjd<dS)N)lengthrversion sequence_r>protein molecule_type) rKrrr)r*rrr1rseq)r6rrr.r#r$_parse_sequences z%Parser.parse.._parse_sequenceNr>)rXdatasetUnknownDatasetrr2 accessionrgenerOr] organismHostkeywordrrWrrrevidencesequencer8)r)rrrrr)rKrr4rrr*sortedsetr3rX)r&r7r:rHrMrRrbrdrfrrrrrrrrrr6r#)rr/r&r$rSsz     0 G                     z Parser.parseNF)__name__ __module__ __qualname____doc__r'rr#r#r#r$rDs  rr)r xml.etreerxml.parsers.expatrBiorBio.Seqr Bio.SeqRecordrrrr%rr#r#r#r$s     &