Fig. 1

A-C Within species guilt-by-association predictive model. D-G Between species structural similarity predictive model. A 20 measures of protein similarity are calculated for all pairs of P. putida proteins. B The depth of deepest shared GO term is predicted for each pair of proteins based on the 20 measures of similarity. Hits are considered to have a depth > 6. C The matching of a term to a protein is predicted from summary statistics of hits that contain the term. D Alphafold structure predictions are downloaded from the Alphafold database. E PUF structures and a matching number of PKF structures are searched against the PDB using RUPEE. F Predictors are calculated from RUPEE structure alignments and NWalign sequence alignments. G Annotation matches are predicted as in (C) from hits annotated with each term. Final annotations are the union of the output of the two models