RNA SEQUENCE-ANALYSIS USING COVARIANCE-MODELS

We describe a general approach to several RNA sequence analysis proble ms using probabilistic models that flexibly describe the secondary str ucture and primary sequence consensus of an RNA sequence family. We ca ll these models 'covariance models'. A covariance model of tRNA sequen ces is an extremely sensitive and discriminative tool for searching fo r additional tRNAs and tRNA-related sequences in sequence databases. A model can be built automatically from an existing sequence alignment. We also describe an algorithm for learning a model and hence a consen sus secondary structure from initially unaligned example sequences and no prior structural information. Models trained on unaligned tRNA exa mples correctly predict tRNA scondary structure and produce high-quali ty multiple alignments. The approach may be applied to any family of s mall RNA sequences.