The
RNA
molecules have long been known for their role in the Central Dogma of molecular
biology. Recently many new kinds of RNA molecules and new biological functions
for RNA molecules have also been found. Given the current massive amount of
sequence being generated, we need to have bioinformatics methods for
recognizing and predicting the structure and function of RNA molecules from
primary sequence data. -- Chomsky transformational grammars, borrowed from
linguistics, have been used as a theoretical basis for modeling certain kinds
of RNA structures. In practice, these methods were unable to predict the
category of RNA structure known as a pseudoknot. This
was frustrating as more and more examples of RNA pseudoknots
have been discovered to have interesting biological functions. -- We have
developed an extension of stochastic context-free grammars that permits them to
be used to computationally predict RNA pseudoknot
structures. We are further developing this method with the goals of:
improving RNA structure and function prediction; and, searching sequence databases for
these kinds of structures to generate testable hypotheses about sequence
function.
Cai, L., Malmberg, R.L., Wu, Y. 2003 Stochastic modeling of RNA pseudoknotted structures : a
grammatical approach. Bioinformatics 19(s1) i66 i73.