A New Approach for Parameter Estimation in the Sequence-structure Alignment of Non-coding RNAs

Recently, searching genomes with a computer program has become an important approach for identifying new noncoding RNAs (ncRNA). Such a computer program often determines whether a sequence segment is the searched ncRNA or not by aligning the sequence segment to a secondary structure model for the searched ncRNA family. To a large extent, the search accuracy depends on the accuracy of the secondary structure model. In this paper, we develop a novel algorithm that can estimate the parameters associated with a few crucial structure features that have been proposed in previous work. This algorithm determines the relative importance of the crucial structure features by solving a convex optimization problem whose objective is to maximize the recognition ability of the structure model. Our experiments also show that this new parameter estimation algorithm can significantly improve the search accuracy