o Rŀg=@sdZddlZgdZgdZgdZgdZgdZgdZgd Zd d Z d d Z ddZ ddZ ddZ ddZddZddZddZddZddZd d!Zd"d#Zefd$d%Zefd&d'Zd(d)Zed*krwdd+lmZedd,dSdS)-aParsers for the GAF, GPA and GPI formats from UniProt-GOA. Uniprot-GOA README + GAF format description: ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/UNIPROT/README Gene Association File, GAF formats: http://geneontology.org/docs/go-annotation-file-gaf-format-2.2/ http://geneontology.org/docs/go-annotation-file-gaf-format-2.1/ http://geneontology.org/docs/go-annotation-file-gaf-format-2.0/ Gene Product Association Data (GPA format) README: http://geneontology.org/docs/gene-product-association-data-gpad-format/ Gene Product Information (GPI format) README: http://geneontology.org/docs/gene-product-information-gpi-format/ Go Annotation files are located here: ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/ N)DB DB_Object_IDDB_Object_Symbol QualifierGO_ID DB:ReferenceEvidenceWithAspectDB_Object_NameSynonymDB_Object_TypeTaxon_IDDate Assigned_ByAnnotation_ExtensionGene_Product_Form_ID)rrrrrrrr r r r r rrr) rrrrrz Evidence coder Interacting_taxon_IDr Assigned_byr Spliceform_ID) rrrrrECO_Evidence_coder rrrzAnnotation ExtensionAnnotation_Properties) r DB_subsetrrr DB_Object_Synonymr TaxonAnnotation_Target_SetAnnotation_CompletedParent_Object_ID) rrr rr rrDB_XrefGene_Product_Properties) rrrr rr rrrrccsp|D]2}|ddkr q|dd}t|dkrq|dd|d<|dd|d<ttt|Vqd S) zRead GPI 1.0 format files (PRIVATE). This iterator is used to read a gp_information.goa_uniprot file which is in the GPI 1.0 format. r!  |N)rstripsplitlendictzip GPI10FIELDShandleinlineinrecr1C/var/www/html/myenv/lib/python3.10/site-packages/Bio/UniProt/GOA.py_gpi10iterators  r3cc|D]D}|ddkr q|dd}t|dkrq|dd|d<|dd|d<|d d|d <|d d|d <ttt|Vqd S) zRead GPI 1.1 format files (PRIVATE). This iterator is used to read a gp_information.goa_uniprot file which is in the GPI 1.1 format. rr r!r"r#r%r&N)r'r(r)r*r+ GPI11FIELDSr-r1r1r2_gpi11iterator  r9ccr4) zRead GPI 1.2 format files (PRIVATE). This iterator is used to read a gp_information.goa_uniprot file which is in the GPI 1.2 format. rr r!r"r#r6r%r& N)r'r(r)r*r+ GPI12FIELDSr-r1r1r2_gpi12iteratorr:r>cCsh|}|dkrt|S|dkrt|S|dkr"t|S|dkr,tdtd|d)zRead GPI format files. This function should be called to read a gp_information.goa_uniprot file. At the moment, there is only one format, but this may change, so this function is a placeholder a future wrapper. z!gpi-version: 1.2z!gpi-version: 1.1z!gpi-version: 1.0z!gpi-version: 2.1z1Sorry, parsing GPI version 2 not implemented yet.zUnknown GPI version r!)readlinestripr>r9r3NotImplementedError ValueErrorr.r/r1r1r2 gpi_iterators    rDccr4) zRead GPA 1.0 format files (PRIVATE). This iterator is used to read a gp_association.* file which is in the GPA 1.0 format. Do not call directly. Rather, use the gpaiterator function. rr r!r"r#r5r%r; N)r'r(r)r*r+ GPA10FIELDSr-r1r1r2_gpa10iterator  rHccr4) zRead GPA 1.1 format files (PRIVATE). This iterator is used to read a gp_association.goa_uniprot file which is in the GPA 1.1 format. Do not call directly. Rather use the gpa_iterator function rr r!r"r#r5r%r;rErFN)r'r(r)r*r+ GPA11FIELDSr-r1r1r2_gpa11iteratorrIrKcCs@|}|dkrt|S|dkrt|Std|d)zRead GPA format files. This function should be called to read a gene_association.goa_uniprot file. Reads the first record and returns a gpa 1.1 or a gpa 1.0 iterator as needed z!gpa-version: 1.1z!gpa-version: 1.0zUnknown GPA version r!)r?r@rKrHrBrCr1r1r2 gpa_iterator s   rLcc|D]M}|ddkr q|dd}t|dkrq|dd|d<|dd|d<|d d|d <|d d|d <|d d|d <ttt|VqdS Nrr r!r"r#r6r%r$r7rF )r'r(r)r*r+ GAF20FIELDSr-r1r1r2_gaf20iterator  rQccrMrN)r'r(r)r*r+ GAF10FIELDSr-r1r1r2_gaf10iterator.rRrTccd}g}|D]m}|ddkrq|dd}t|dkrq|dd|d<|dd|d<|d d|d <|d d|d <|d d|d <ttt|}|d |krk|rkt|}|g}|d }|Vq|d }||qdS Nrr r!r"r#r6r%r$r7rFrOr)r'r(r)r*r+rScopyappendr.cur_id id_rec_listr/r0cur_recret_listr1r1r2_gaf10byproteiniterator=.    r^ccrUrV)r'r(r)r*r+rPrWrXrYr1r1r2_gaf20byproteiniteratorVr_r`cCsh|}|dkrt|S|dkrt|S|dkr"t|S|dkr,t|Std|d)aIterate over records in a gene association file. Returns a list of all consecutive records with the same DB_Object_ID This function should be called to read a gene_association.goa_uniprot file. Reads the first record and returns a gaf 2.0 or a gaf 1.0 iterator as needed 2016-04-09: added GAF 2.1 iterator & fixed bug in iterator assignment In the meantime GAF 2.1 uses the GAF 2.0 iterator !gaf-version: 2.0!gaf-version: 1.0!gaf-version: 2.1!gaf-version: 2.2Unknown GAF version r!)r?r@r`r^rBrCr1r1r2gafbyproteiniteratoros    rfcCsh|}|dkrt|S|dkrt|S|dkr"t|S|dkr,t|Std|d)a Iterate over a GAF 1.0 or 2.x file. This function should be called to read a gene_association.goa_uniprot file. Reads the first record and returns a gaf 2.x or a gaf 1.0 iterator as needed Example: open, read, interat and filter results. Original data file has been trimmed to ~600 rows. Original source ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/YEAST/goa_yeast.gaf.gz >>> from Bio.UniProt.GOA import gafiterator, record_has >>> Evidence = {'Evidence': set(['ND'])} >>> Synonym = {'Synonym': set(['YA19A_YEAST', 'YAL019W-A'])} >>> Taxon_ID = {'Taxon_ID': set(['taxon:559292'])} >>> with open('UniProt/goa_yeast.gaf', 'r') as handle: ... for rec in gafiterator(handle): ... if record_has(rec, Taxon_ID) and record_has(rec, Evidence) and record_has(rec, Synonym): ... for key in ('DB_Object_Name', 'Evidence', 'Synonym', 'Taxon_ID'): ... print(rec[key]) ... Putative uncharacterized protein YAL019W-A ND ['YA19A_YEAST', 'YAL019W-A'] ['taxon:559292'] Putative uncharacterized protein YAL019W-A ND ['YA19A_YEAST', 'YAL019W-A'] ['taxon:559292'] Putative uncharacterized protein YAL019W-A ND ['YA19A_YEAST', 'YAL019W-A'] ['taxon:559292'] rarcrdrbrer!)r?r@rQrTrBrCr1r1r2 gafiterators%    rgcCsd}|ddD]'}t||tr'||D]}||d7}q|ddd}q|||d7}q|||dd7}||dS)zWrite a single UniProt-GOA record to an output stream. Caller should know the format version. Default: gaf-2.0 If header has a value, then it is assumed this is the first record, a header is written. Nr%r"r!) isinstancelistwrite)outrecr.fieldsoutstrfieldsubfieldr1r1r2writerecs rrcCs|D] }t|||dqdS)aOWrite a list of GAF records to an output stream. Caller should know the format version. Default: gaf-2.0 If header has a value, then it is assumed this is the first record, a header is written. Typically the list is the one read by fafbyproteinrec, which contains all consecutive lines with the same DB_Object_ID )rnN)rr) outprotrecr.rnrmr1r1r2writebyproteinrecsrtcCsNd}|D] }t||tr||h}nt||}|||@r$d}|Sq|S)zAccept a record, and a dictionary of field values. The format is {'field_name': set([val1, val2])}. If any field in the record has a matching value, the function returns True. Otherwise, returns False. FT)rjstrset)r0 fieldvalsretvalrpset1r1r1r2 record_hass   rz__main__) run_doctest)verbose)__doc__rWrPrSrGrJr,r8r=r3r9r>rDrHrKrLrQrTr^r`rfrgrrrtrz__name__ Bio._utilsr|r1r1r1r2s:  :