The GOtcha method
The GOtcha method was described in detail in Martin et al. BMC Bioinformatics (2004) 5:178. In summary, GOtcha assigns functional terms transitively based upon sequence similarity. The detailed method is outlined in the figure below.

  1. GOtcha performs a BLAST search of the query sequence against individual well annotated genomes.
  2. Annotations are transitively assigned from all hits, with a score corresponding to the E-value, individual GO-terms receiving cumulative scores from multiple sequence similarity matches.
  3. Cumulative scores are normalised and, for each term, two scores are obtained - the I-score which is normalised to the root node, and the C-score which is the cumulative score at the root node.
  4. For each GO-term a precomputed scoring table is used to establish the assignment likelihood for that term given that I-score and that C-score. This is represented as a probability

Scores on each node are assigned and accumulated as shown in the figure below.

Graphical output is provided using the Graphviz library, to provide a heat mapped graph of the assigned GO terms, colour coded by probability.