Running title Covariance models of RNA RNA Sequence Analysis Using Covariance Models.pdfVIP

Running title Covariance mo dels of RNA RNA Sequence Analysis Using Covariance Mo dels Sean R Eddy and Richard Durbin MRC Lab oratory of Molecular Biology Hills Road Cambridge CB QH England sremrclmbcam acuk rdmrclmbcamacuk April Abstract We describ e a general approach to several RNA sequence analysis problems using prob abilistic mo dels that exibly describ e the secondary structure and primary sequence con sensus of an RNA sequence family We call these mo dels covariance mo dels A covariance mo del of tRNA sequences is an extremely sensitive and discriminative to ol for searching for additional tRNAs and tRNArelated sequences in sequence databases A mo del can b e built automatically from an existing sequence alignment We also describ e an algorithm for learning a mo del and hence a consensus secondary structure from initially unaligned example sequences and no prior structural information Mo dels trained on unaligned tRNA examples correctly predict tRNA secondary structure and pro duce highquality multiple alignments The approach may b e applied to any family of small RNA sequences Sub ject Computational biology Keywords tRNA covariance mo del sto chastic contextfree grammar database searching multiple alignment RNA secondary structure Intro duction A ma jor role of computational metho ds in molecular biology is to identify similarities b etween sequences Similarity b etween sequences generally implies functional andor evo lutionary homology and therefore provides imp ortant biological information The analysis of largescale genome sequence data is particularl