Modeling Structural RNA Families Using Covariance Models

• Activity Number: 630
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Eric Nawrocki*+
• Companies: Howard Hughes Medical Institute
• Keywords: RNA ; sequence analysis ; probabilistic models

Abstract

Most genes encode proteins: DNA is transcribed to messenger RNA and then translated to protein, which carries out a biochemical function in the cell. Some RNAs, however, do not encode proteins but rather function directly as RNAs. It is useful to group together evolutionarily-related, or homologous, RNAs into families and compare them, as it often leads to functional insight. Many of these RNAs form an evolutionarily-conserved structure and include sequence features that are crucial for their function. However, relative to protein-coding genes, structural RNA genes are difficult to computationally identify in genome sequences.

A covariance model (CM) is a probabilistic model of the sequence and structure of an RNA family built from a set of known examples. The Rfam database includes CMs for more than 2000 RNA families. The Infernal software package implements dynamic programming algorithms for calculating a log-odds score that a subsequence was generated from a CM, a model of homology for the family, versus a background model of non-homology. Infernal is used in conjunction with the Rfam database to reliably identify structural RNAs from known families in sequence databases.