ARE KNOWLEDGE-BASED POTENTIALS DERIVED FROM PROTEIN-STRUCTURE SETS DISCRIMINATIVE WITH RESPECT TO AMINO-ACID TYPES

The parametric description of residue environments through solvent acc essibility, backbone conformation, or pairwise residue-residue distanc es is the key to the comparison between amino acid types at protein se quence positions and residue locations in structural templates (condit ion of protein sequence-structure match). For the first time, the rese arch results presented in this study clarify and allow to quantify, on a rigorous statistical basis, to what extent the amino acid type-spec ific distributions of commonly used environment parameters are discrim inative with respect to the 20 amino acid types. Relying on the Bahadu r theory, we estimate the probability of error in a single-sequence-st ructure alignment based on weak or absent discriminative power in a le arning database of protein structure. We present the results for many residue environment variables and demonstrate that each fold descripti on parameter is sensitive with respect to only a few amino acid types while indifferent to most of the other amino acid types, Even complex structural characteristics combining solvent-accessible surface area, backbone conformation, and pairwise distances distinguish only some am ino acid types, whereas the others remain nondiscriminated. We find th at the knowledge-based potentials currently in use treat especially Al a, Asp, Gin, His, Ser, Thr, and Tyr as essentially ''average'' amino a cids. Thus, highly discriminative amino acid types define the alignmen t register in gapless sequence-structure alignments. The introduction of gaps leads to alignment ambiguities at sequence positions occupied by nondiscriminated amino acid types. Therefore, local sequence-struct ure alignments produced by techniques with gaps cannot be reliable. Co nceptionally new and more sensitive environment parameters must be inv ented. (C) 1998 Wiley-Liss, Inc.