PROSPECT: A Computer System for Globally-Optimal Threading
This paper presents a new computer system, PROSPECT, for protein threading. PROSPECT employs an energy function that consists of three additive terms: (1) a singleton fitness term, (2) a distance-dependent pairwise-interaction preference term, and (3) alignment gap penalty; and currently uses FSSP as its threading template database. PROSPECT uses a divide-and-conquer algorithm to find an alignment between a query protein sequence and a protein fold template, which is guaranteed to be globally optimal for its energy function. The threading algorithm presented here significantly improves the computational efficiency of our previously-published algorithm, which makes PROSPECT a practical tool even for large protein threading problems. Mathematically, PROSPECT finds a globally-optimal threading between a query sequence of n residues and a fold template of m residues and M core secondary structures in O(nm + MnN{sup 1.5C{minus}1}) time and O(nm + nN{sup C{minus}1}) space, where C, the topological complexity of the template fold as we term, is a value which characterizes the overall structure of the considered pairwise interactions in the fold; and N represents the maximum number of possible alignments between an individual core of the fold and the query sequence when its neighboring cores are already aligned. PROSPECT allows a user to incorporate known biological constraints about the query sequence during the threading process. For given constraints, the system finds a globally-optimal threading which satisfies the constraints. Currently PROSPECT can deal with constraints which reflect geometrical relationships among residues of disulfide bonds, active sites, or determined by the NOE constraints of (low-resolution) NMR spectral data.
- Research Organization:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- USDOE Office of Energy Research (ER) (US)
- DOE Contract Number:
- AC05-96OR22464
- OSTI ID:
- 3929
- Report Number(s):
- ORNL/CP-101519; KP 11 01 01 0; KP 11 01 01 0; TRN: AH200113%%182
- Resource Relation:
- Conference: 7th International Conference on Intelligent Systems for Molecular Biology (ISMB '99), Heidelberg (DE), 08/06/1999; Other Information: PBD: 6 Aug 1999
- Country of Publication:
- United States
- Language:
- English
Similar Records
Protein Threading by PROSPECT: A Prediction Experiment in CASP3
Protein Structure Determination Using Protein Threading and Sparse NMR Data