:>"^6@s!dZddlmZddlZddlmZddddd d d d d ddddddddgZddddd d d d d ddddddgZdddd d dd dddddg Zdddd d dd dddddg Z ddddddddd d!d"g Z ddddddd"d#d$g Z d%d&Z d'd(Z d)d*Zd+d,Zd-d.Zd/d0Zd1d2Zd3d4Zd5d6Zd7d8Zd9d:Zd;d<Zed=d>Zed?d@ZdAdBZedCkrddDlmZedEddS)FaParsers 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 GAF formats: http://geneontology.org/docs/go-annotation-file-gaf-format-2.1/ http://geneontology.org/docs/go-annotation-file-gaf-format-2.0/ gp_association (GPA format) README: http://geneontology.org/docs/gene-product-association-data-gpad-format/ gp_information (GPI format) README: http://geneontology.org/docs/gene-product-information-gpi-format/ )print_functionN)zipZDB DB_Object_IDZDB_Object_SymbolZ QualifierZGO_IDz DB:ReferenceZEvidenceWithZAspectZDB_Object_NameZSynonymZDB_Object_TypeZTaxon_IDDateZ Assigned_ByZAnnotation_ExtensionZGene_Product_Form_IDz Evidence codeZInteracting_taxon_IDZ Assigned_byZ Spliceform_IDZECO_Evidence_codezAnnotation ExtensionZAnnotation_PropertiesZ DB_subsetZDB_Object_SynonymZTaxonZAnnotation_Target_SetZAnnotation_CompletedZParent_Object_IDZDB_XrefZGene_Product_Propertiesccsx|D]}|ddkr q|jdjd}t|dkrMq|djd|d<|djd|d>> 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'] z!gaf-version: 2.0z!gaf-version: 2.1z!gaf-version: 1.0zUnknown GAF version {0} N)rrr.r0r!r")rrrrr gafiteratords%    r:cCsd}xw|ddD]e}t||trjx ||D]}||d7}q;W|dd d}q|||d7}qW|||d d7}|jd|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 rz%sr<r<) isinstancelistwrite)outrecrfieldsZoutstrfieldZsubfieldrrrwriterecsrCcCs(x!|D]}t||d|qWdS)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 rAN)rC)Z outprotrecrrAr@rrrwritebyproteinrecs rDcCscd}xV|D]N}t||tr6||h}nt||}|||@r d}Pq W|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)r=strset)rZ fieldvalsretvalrBZset1rrr record_hass rH__main__) run_doctestverbose)__doc__ __future__rr1Z Bio._py3krr-r/r'r)rrrrr#r(r*r+r.r0r7r8r9r:rCrDrH__name__Z Bio._utilsrJrrrrs                   4