Promoter recognition based on the Interpolated Markov Chains optimized via simulated annealing and genetic algorithm

Simulated annealing (SA) and genetic algorithm (GA) are utilized to optimize Interpolation Markov Chains (IMC) model for promoter recognition. The deletions and insertions of nucleotides in DNA sequences are introduced into the IMC model whose transition probabilities are established with SA. The noise is filtered to reduce the complexity of model parameters. And to improve the gradient descent algorithm being liable to fall into the local minimum point, GA is presented for an automated estimation of global optimal interpolation coefficients. A simulation result shows that the sensitivity and specificity in promoter level are both higher than 86% on the test set.