GO filtering

We have a priority list of evidence codes. Currently:

inferred from biological aspect of descendant
inferred from biological aspect of ancestor
inferred from reviewed computational analysis
inferred from electronic annotation
inferred from expression pattern
non-traceable author statement
traceable author statement
inferred from sequence model
inferred from sequence or structural similarity
inferred from sequence orthology
inferred by curator

Annotations with evidence codes at the bottom are kept in preference to those nearer the top of the list.

• we iterate through the evidence codes above in order then:

  • for each annotation with that evidence code, delete the annotation if there is another annotation for the same gene with:

    • a more specific term
    • or same term but a different evidence code

  • if there are duplicate annotations (same term and gene) with that evidence code, delete all but one