The evaluation of a stochastic regular motif language for protein sequences
Brian J. Ross · 2001
A probabilistic regular motif language for protein sequences is evaluated. SRE-DNA is a stochastic regular expression language that combines characteristics of regular expressions and stochastic representations such as Hidden Markov Models. To evaluate its expressive merits, genetic programming is used to evolve SRE-DNA motifs for aligned sets of protein sequences. Di#erent constrained grammatical forms of SRE-DNA expressions are applied to aligned protein sequences from the PROSITE database. Some sequences patterns were precisely determined, while others resulted in good solutions having considerably di#erent features from the PROSITE equivalents. This research establishes the viability of SRE-DNA as a new representation language for protein sequence identification. The practicality of using grammatical genetic programming in stochastic biosequence expression classification is also demonstrated.