Importing GO terms from UniProt to a PGDB

1
Importing GO terms from UniProt to a PGDB

• Markus Krummenacker
• Bioinformatics Research Group
• SRI International
• kr_at_ai.sri.com

2
GO in EcoCyc Introduction

• GO (http//geneontology.org) is used widely to
  annotate gene products with functions, processes,
  and cellular locations
• Manual curation of GO annotations in EcoCyc

3
UniProtKB GO annotations

• GO consortium hosts UniProtKB annotations file
• Big, several GB. grep file for E. coli taxon
  ID
• Import code maps UniProtKB IDs to EcoCyc gene
  products, via DBLINKs of the products
• Most imported GO annots have comp. evidence
• Comp. ev. annots get timestamps bumped up
  (because they expire after 1 yr.)
• Suppress comp. ev. annots if redundant with an
  existing exp. ev. annot
• Prune comp. ev. annots if a more specific annot
  of the same kind exists (several dozens)

4
EcoliWiki EcoCyc collaboration

• Collaboration with Jim Hu / EcoliWiki
• Workflow
• GO UniProtKB ? EcoCyc
• EcoCyc exports GO annots file
• EcoCyc GO annots ? EcoliWiki
• Merging of EcoCyc and additional EcoliWiki annots
• EcoliWiki ? GO consortium, deposit file for E.
  coli
• Annots are absorbed into UniProtKB
• Repeat in half a year

5
Open Issues

• Round-trip problem of deleted annots
• EcoCyc curator deletes an annot, because wrong
• EcoliWiki should detect this. Protocol not clear
  yet.
• For now UniProtKB import into EcoCyc checks
  history logs, to prevent annot addition if that
  annot was deleted in the past
• No EcoCyc support yet for some qualifiers
• NOT
• Contributes_to
• No easy user interface yet for annot import

6
Do it Yourself

• Disclaimer Has never been tried outside of
  EcoCyc
• Prepare input file (using grep). DBLINKS need to
  exist on gene products.
• (add-go-terms-to-monomers
• (incorporate-ecocyc-go-terms-from-GOAFF-f
  ile
• filename ../gene_association.goa
• db-type UNIPROT) )
• (save-kb)
• (loop for p in (all-frames-that-could-contain-go
  -annots)
• do (prune-unnecessary-go-terms p
  destructively-prune! t))