B bq@sJdZddlmZddlmZddlmZddlmZd ddZdd Z d S) a?Bio.SeqIO support for the "swiss" (aka SwissProt/UniProt) file format. You are expected to use this module via the Bio.SeqIO functions. See also the Bio.SwissProt module which offers more than just accessing the sequences as SeqRecord objects. See also Bio.SeqIO.UniprotIO.py which supports the "uniprot-xml" format. ) SeqFeature) SwissProt)Seq) SeqRecordc Cs|dkrtSyttd|t|Stk r<YnX|dry ttd|t|ddStk r|YnXn|dry ttd|t|ddStk rYnXnD|dry t td|t|ddStk rYnXt d|dS)zTurn a Swiss location position into a SeqFeature position object (PRIVATE). An offset of -1 is used with a start location to make it pythonic. ?r<N>zCannot parse location '%s') rZUnknownPositionZ ExactPositionmaxint ValueError startswithZBeforePositionZ AfterPositionZUncertainPositionNotImplementedError)Zlocation_stringoffsetr0lib/python3.7/site-packages/Bio/SeqIO/SwissIO.py_make_positions.     rccspt|}x^|D]T}tt|j|jd|j|j|jd}xN|j D]D}t |dkrVqD|dd\}}d||f}||j krD|j |qDW|j }d|d<|j|d<|jr|j|d <|jr|j\} } | |d <| |d <|jr|j\} } | |d <| |d <|jr|j\} } | |d <| |d<|jr(|j|d<|jd|d<|j|d<|j|d<|jr^|j|d<|jrp|j|d<|jrd|j|d<|jrNg|d<x|jD]} t} ddd| jD| _x\| jD]R\} }| dkr|| _n6| dkr|| _ n$| dkrn| dkrn t!d | qW| j"| _"| j#| _#| j$| _%|d | qW|j&rb|j&|j d!<|VqWdS)"aBreak up a Swiss-Prot/UniProt file into SeqRecord objects. Argument source is a file-like object or a path to a file. Every section from the ID line to the terminating // becomes a single SeqRecord with associated annotation and features. This parser is for the flat file "swiss" format as used by: - Swiss-Prot aka SwissProt - TrEMBL - UniProtKB aka UniProt Knowledgebase For consistency with BioPerl and EMBOSS we call this the "swiss" format. See also the SeqIO support for "uniprot-xml" format. Rather than calling it directly, you are expected to use this parser via Bio.SeqIO.parse(..., format="swiss") instead. r)idname descriptionfeaturesNz%s:%sZproteinZ molecule_type accessionsprotein_existencedateZsequence_versionZdate_last_sequence_updateZdate_last_annotation_updateZ entry_version gene_name.organismZtaxonomyZ ncbi_taxidZ organism_hostZhost_ncbi_taxid comment references css|]}d|VqdS)z%s=%s;Nr).0Zk_vrrr |sz SwissIterator..ZPubMedZMEDLINEZDOIZAGRICOLAz"Unknown key %s found in referenceskeywords)'rparserrZsequencerZ entry_namerrZcross_referenceslenZdbxrefsappend annotationsrZcreatedZsequence_updateZannotation_updaterrrstripZorganism_classificationZ taxonomy_idZ host_organismZhost_taxonomy_idZcommentsjoinr rZ ReferencerZ pubmed_idZ medline_idr ZauthorstitlelocationZjournalr$)sourceZ swiss_recordsZ swiss_recordrecordZcross_referenceZdatabaseZ accessionZdbxrefr(rversionZ referenceZfeaturekeyvaluerrr SwissIterator7s                      r2N)r) __doc__ZBiorrZBio.SeqrZ Bio.SeqRecordrrr2rrrrs