Xf=dZddlZgdZgdZgdZgdZgdZgdZgd Zd Z d Z d Z d Z dZ dZdZdZdZdZdZdZdZefdZefdZdZedkrddlmZeddSdS)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)rrrrrrr r r r r rrrr) rrrrrz Evidence coder Interacting_taxon_IDr Assigned_byr Spliceform_ID) rrrrrECO_Evidence_coder rrrzAnnotation ExtensionAnnotation_Properties) r DB_subsetrrr DB_Object_SynonymrTaxonAnnotation_Target_SetAnnotation_CompletedParent_Object_ID) rrr rrrrDB_XrefGene_Product_Properties) rrrr rrrrrr c#fK|D]}|ddkr|dd}t|dkrK|dd|d<|dd|d<tt t |Vd 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 GPI10FIELDShandleinlineinrecs /lib/python3.11/site-packages/Bio/UniProt/GOA.py_gpi10iteratorr4s ,, !9    d##))$// u::?? 8>>#&&a8>>#&&a3{E**++++++,,c#K|D]}|ddkr|dd}t|dkrK|dd|d<|dd|d<|d d|d <|d d|d <tt t |Vd S) zRead GPI 1.1 format files (PRIVATE). This iterator is used to read a gp_information.goa_uniprot file which is in the GPI 1.1 format. rr"r#r$r%r'r(N)r)r*r+r,r- GPI11FIELDSr/s r3_gpi11iteratorr;  , , !9    d##))$// u::?? 8>>#&&a8>>#&&a8>>#&&a8>>#&&a3{E**++++++ , ,r5c#K|D]}|ddkr|dd}t|dkrK|dd|d<|dd|d<|d d|d <|d d|d <tt t |Vd S) zRead GPI 1.2 format files (PRIVATE). This iterator is used to read a gp_information.goa_uniprot file which is in the GPI 1.2 format. rr"r#r$r%r8r'r( N)r)r*r+r,r- GPI12FIELDSr/s r3_gpi12iteratorrAr<r5c|}|dkrt|S|dkrt|S|dkrt |S|dkrt dt d|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#)readlinestriprAr;r4NotImplementedError ValueErrorr0r1s r3 gpi_iteratorrHs__  F ||~~,,,f%%% . . .f%%% . . .f%%% . . .""UVVV::::;;;r5c#K|D]}|ddkr|dd}t|dkrK|dd|d<|dd|d<|d d|d <|d d|d <tt t |Vd S) 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$r%r7r'r> N)r)r*r+r,r- GPA10FIELDSr/s r3_gpa10iteratorrM , , !9    d##))$// u::?? 8>>#&&a8>>#&&a8>>#&&a"IOOC((b 3{E**++++++ , ,r5c#K|D]}|ddkr|dd}t|dkrK|dd|d<|dd|d<|d d|d <|d d|d <tt t |Vd S) 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$r%r7r'r>rJrKN)r)r*r+r,r- GPA11FIELDSr/s r3_gpa11iteratorrQrNr5c|}|dkrt|S|dkrt|St d|d)zRead GPA format files. This function should be called to read a gene_association.goa_uniprot file. Reads the first record and returns a gpa 1.1 or a gpa 1.0 iterator as needed z!gpa-version: 1.1z!gpa-version: 1.0zUnknown GPA version r#)rCrDrQrMrFrGs r3 gpa_iteratorrS sq__  F ||~~,,,f%%% . . .f%%%::::;;;r5c#K|D]}|ddkr|dd}t|dkrL|dd|d<|dd|d<|d d|d <|d d|d <|d d|d <tt t |VdS Nrr"r#r$r%r8r'r&r9rK )r)r*r+r,r- GAF20FIELDSr/s r3_gaf20iteratorrX , , !9    d##))$// u::?? 8>>#&&a8>>#&&a8>>#&&a"IOOC((b "IOOC((b 3{E**++++++ , ,r5c#K|D]}|ddkr|dd}t|dkrL|dd|d<|dd|d<|d d|d <|d d|d <|d d|d <tt t |VdSrU)r)r*r+r,r- GAF10FIELDSr/s r3_gaf10iteratorr\.rYr5c#Kd}g}|D]T}|ddkr|dd}t|dkrL|dd|d<|dd|d<|d d|d <|d d|d <|d d|d <tt t |}|d |kr'|r%t j|}|g}|d }|V7|d }||VdS Nrr"r#r$r%r8r'r&r9rKrVr)r)r*r+r,r-r[copyappendr0cur_id id_rec_listr1r2cur_recret_lists r3_gaf10byproteiniteratorrf=a FK(( !9    d##))$// u::?? 8>>#&&a8>>#&&a8>>#&&a"IOOC((b "IOOC((b s;..// > "f , , ,y--H")K^,FNNNN^,F   w ' ' ' ''((r5c#Kd}g}|D]T}|ddkr|dd}t|dkrL|dd|d<|dd|d<|d d|d <|d d|d <|d d|d <tt t |}|d |kr'|r%t j|}|g}|d }|V7|d }||VdSr^)r)r*r+r,r-rWr_r`ras r3_gaf20byproteiniteratorriVrgr5c|}|dkrt|S|dkrt|S|dkrt|S|dkrt|St d|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#)rCrDrirfrFrGs r3gafbyproteiniteratorrpos__  F ||~~,,,&v... . . .&v... . . .'v... . . . 'v...::::;;;r5c|}|dkrt|S|dkrt|S|dkrt|S|dkrt|St d|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'] rkrmrnrlror#)rCrDrXr\rFrGs r3 gafiteratorrrsJ__  F ||~~,,,f%%% . . .f%%% . . . f%%% . . .f%%%::::;;;r5c d}|ddD]L}t||tr!||D] }||dzz } |dddz}>|||dzz }M|||ddzz }||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)outrecr0fieldsoutstrfieldsubfields r3writerecr~sF++ fUmT * * +"5M ) )(S.(CRC[4'FF fUmd* *FF fVBZ 4''F LLr5c4|D]}t|||dS)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 )rzN)r~) outprotrecr0rzrys r3writebyproteinrecrs500/////00r5cd}|D]K}t||tr ||h}nt||}|||zrd}nL|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)rvstrset)r2 fieldvalsretvalr|set1s r3 record_hasrstF eElC ( ( %%L>DDuU|$$D )E" " F E  Mr5__main__) run_doctest)verbose)__doc__r_rWr[rLrPr.r:r@r4r;rArHrMrQrSrXr\rfrirprrr~rr__name__ Bio._utilsrr5r3rs(  * &                  ,,,",,,&,,,&<<<2,,,(,,,(<<<$ , , , , , ,(((2(((2<<<>7<7<7