Author's School

Graduate School of Arts & Sciences

Author's Department/Program

Biology and Biomedical Sciences: Computational and Systems Biology

Language

English (en)

Date of Award

January 2010

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Chair and Committee

Sean Eddy

Abstract

Non-coding RNAs are biologically important molecules, with a variety of catalytic and regulatory activities mediated by their secondary and tertiary structures. Base-pairing interactions, particularly at the secondary structure level, are an important tool for identifying and studying these structural RNAs, but also present some unique challenges for sequence analysis. Probabilistic covariance models are effective representations of structural RNAs, with generally high sensitivity and specificity but slow computational speed. New algorithms for dealing with structural RNAs are developed to address some of the practical deficiencies of covariance models. A new model of local alignment improves accuracy for fragmentary data, such as found in direct shotgun sequencing and metagenomic surveys. A separate and alternative model of local alignment is used as the basis for a structural search filter. This is combined with other filtering techniques and high-performance implementations to increase the practical speed of high-sensitivity search. As a whole, these improvements provide a foundation for and point toward future improvements in noncoding RNA homology search.

Comments

Permanent URL: http://dx.doi.org/10.7936/K7CR5RC5

Download

Share

COinS