Hyper-redundant Robots and bioinformatics: Modelling loops in RNA

Abstract

Nucleic acid structures, like DNA and RNA, primarily involve hydrogen bonded, base-paired stretches interspersed with loops of various geometric shapes. Theoretical modelling and simulation of these loop structures have been an analytical challenge (with the exception of proteins) and most methods used are based on heuristics. The objectives of this research are to model loops in nucleic acid structures by developing an algorithm using the inverse kinematics for hyper-redundant manipulators and develop a database of theoretical analogue for corresponding experimental models in Protein Data Bank (PDB). First, a volumetric representation of the loops in nucleic acids is sought from a hyper-redundant manipulator schema with torsion angles as joint variables. Based on it, steric clashes, as a function of loop configuration, in side chains are predicted, acknowledging that a stable configuration should have minimal infringement. It is further shown that loop modelling is essentially a process of constrained energy optimization for redundant structures which, at its background, requires kinematic approximation of the loop and estimation of infringement. Simulated theoretical prediction of structure, of some such loops in specific RNA molecules and non-Watson-Crick duplex or quadruplex (along with statistical parameters), illustrate the validity of this approach to modelling. Predicted structures (with multiple possibilities) are considered for selection only if certain statistical criteria (indicating proximity) are satisfied when compared to their experimental counterparts.