Identifying amino acid residues in medium resolution critical point graphs using instance based query generation

Instance Based Query Generation is defined and applied to the problem of recognising amino acid residues in medium resolution critical point graphs. The technique is an amalgamation of Relational Instance Based Learning and Frequent Query Discovery in First Order Logic. Instances are automatically constructed from a deductive database and first order association rules are derived from the instances. The initial investigations presented here indicate that the technique is able to discriminate some of the larger amino acid types as well as discriminating the protein from background solvent. Identification of the smaller amino acids remains difficult and requires further work.