Skip to Main Content
Systematic research on noncoding RNAs (ncRNAs) has revealed that many ncRNAs are actively involved in various biological networks. Therefore, in order to fully understand the mechanisms of these networks, it is crucial to understand the roles of ncRNAs. Unfortunately, the annotation of ncRNA genes that give rise to functional RNA molecules has begun only recently, and it is far from being complete. Considering the huge amount of genome sequence data, we need efficient computational methods for finding ncRNA genes. One effective way of finding ncRNA genes is to look for regions that are similar to known ncRNA genes. As many ncRNAs have well-conserved secondary structures, we need statistical models that can represent such structures for this purpose. In this paper, we propose a new method for representing RNA sequence profiles and finding structural alignment of RNAs based on profile context-sensitive hidden Markov models (profile-csHMMs). Unlike existing models, the proposed approach can handle any kind of RNA secondary structures, including pseudoknots. We show that profile-csHMMs can provide an effective framework for the computational analysis of RNAs and the identification of ncRNA genes.