Abstract:A novel model of genomic data mining and a corresponding algorithm for pattern discovery were proposed . The model consists of six units such as artificial genome, artificial proteome, evolutionary operation, evolutionary control, pattern matching and termination judgement. The abstract algebraic structure is described by lattice set configuration and finite state automata dynamically. The candidate string sequence is generated by evolutionary algorithm with symbolic dynamics. The degree of evolution is controlled by meta-evolution mechanism and expressed by rough sets. The pattern matching procedure is implemented by syntatic pattern recognizer and grammar inference. Termination judgement is dependent on concret problem solving paradigm. The algorithm is with the cycle type of implicit parallation and population searching . The data structure focusses on coarse-grained symbolic information processing and modular programming oriented to semantics. With the application of the above-mentioned artificial life techniques, candidate symbolic sequences have been automatically produced by computer system and “ real” amino-acid sequence obtained among them. The experimental result shows that the computational method proposed and implemented here is helpful to the building of unified computational theory of bioinformatics in the genomics level and development of application systems.

Key words:         genomics         bioinformatics         evolutionary computation