Efficient Tree Grammatical Modeling of RNA Secondary Structures from Alignment Data

Recent advances of research on non-coding RNAs (nc-RNAs) reveal that they are more numerous and important than previously thought ([8][5], etc.). In order to find new members of nc-RNA families, there have been proposed various powerful grammatical tools which can model both primary sequences and secondary structure information ([7][4][9][2], etc.). In order to develop a full automatic grammatical system for finding new members of nc-RNA families, we select a tree adjoining grammar (TAG) as a model of RNAs. In this manuscript, we will give an efficient linear time algorithm which constructs probabilistic tree adjoining grammar (PTAG) models of RNA secondary structures from alignment data.