Statement of Problem:
Given an RNA target, several hurdles exist to identifying small molecule binders and their binding sites because molecular recognition involves both sequence and structure. We propose to address these hurtles by developing new algorithms and a software program, which we call RNAbind. As input, RNAbind will take the sequence of the target sequence and a database of suitable RNA binding molecules. RNAbind will then output the set of molecules that are predicted to have binding potential and the probability that each of the target structural motifs occurs in the structure of the target sequence. Furthermore, RNAbind will assemble a list of potential compound molecules, composed of two RNA binding molecules from the database, based on proximity of the individual binding motifs in the target structure.
RNA secondary structure prediction has traditionally been accomplished by predicting the lowest free energy structure (1), where folding free energy change is evaluated using a nearest neighbor model (2,3). The Mathews lab developed new algorithms that utilize partition functions to predict probabilities of structures and motifs in the complete folding ensemble (4,5). We demonstrated the utility of these approaches in several contexts. For example, we demonstrated the ensemble approach improves the prediction of siRNA and antisense DNA oligonucleotide target sites in mRNA (6,7).
Our first application will be the design of small molecules to target HIV. This will utilize the library of RNA-binding molecules developed by Matthew Disney, State University of New York, Buffalo, and the recent mapping of the HIV structure by Kevin Weeks, University of North Carolina, Chapel Hill (8,9).