Threading with environment-specific score by artificial neural networks

  • Authors:

  • N. Jiang;W. XinyuWu;I. Mitchell

  • Affiliations:

  • School of Computing Science, Middlesex University, The Burroughs, NW4 4BT, London, Hendon;School of Computing Science, Middlesex University, The Burroughs, NW4 4BT, London, Hendon;School of Computing Science, Middlesex University, The Burroughs, NW4 4BT, London, Hendon

  • Venue:
  • Soft Computing - A Fusion of Foundations, Methodologies and Applications
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Protein threading programs align a probe amino acid sequence onto a library of representative folds of known protein structure to identify a structural homology. A scoring function is usually formulated in terms of the threading energy to evaluate the protein sequence-structure fitness. In this paper, a model named threading with environment-specific score (TES) is proposed to build a new threading score function with the use of artificial neural networks. Given a protein structure with a residue level environment description, the compatibility of residue in sequence with its structural environment is presented. A threading score is constructed by log-odds scores of predicted probabilities from the trained model to determine which residue best fits its environment. Two decoy sets are used to test the proposed TES method on discrimination of native and decoy protein three-dimensional structure. The results showed that the performance of the proposed method is comparable to those of knowledge-based potential energy function.