Our work mainly focuses on two broad subjects: Computational methods for RNA structure prediction including the identification of new RNA functions and the integration of phylogenetic methods into commonly used homology search and alignment methods. The high level objective of both aspects of my work is to design new algorithms for reliably determining the nature of any piece of biological sequence. The main tool we use to carry this program forward is the use of probabilistic models and rigorous statistical inference.
RNA computational biology has many urgent issues, because of the discovery of large numbers of long coding RNAs (lncRNAs), whose functional significance remains unclear. We suspect that lncRNAs are heterogeneous, and that many are transcriptional noise and artifacts, but that a subset are functional and important. Conserved RNA structure prediction analysis will be an important tool in understanding functional RNAs, and distinguishing them from artifacts and noise.
To improve the identification of RNA function, we follow at least three directions: (1) find better models of RNA secondary structure prediction so that we can incorporate prior information on structure, and (2) develop better models of evolution so we can use multiple sequences effectively, and (3) we are currently investigating the detection of phylogenetically conserved structures by assessing significant covarying pairs in RNA alignments.