(A) Authorship bias: average GO Similarity of homologs pairs partitioned according to their provenance. (B) Variation of frequencies of GO terms among the 13 analyzed genomes (50 most common terms on average depicted). (C) Average background frequency for the different subtypes of gene pairs, obtained by computing the average similarity of random pairs from sequences involved in the respective categories. (D) Average GO similarity between homologous gene pairs partitioned according to their GO annotation evidence tags (Experimental: evidence code EXP and children; Uncurated: evidence code IEA; Curated: all other evidence codes). To compute the average similarity for each category, annotations from the other 2 categories are filtered out.