We have developed a method for predicting the structure of small RNA l
oops that can be used to augment already existing RNA modeling techniq
ues. The method requires no input constraints on loop configuration ot
her than end-to-end distance. Initial loop structure are generated by
randomizing the torsion angles, beginning at one end of the polynucleo
tide chain and correlating each successive angle with the previous. Th
e bond lengths of these structures are then scaled to fit within the k
nown end constraints and the equilibrium bond lengths of the potential
energy function are scaled accordingly. Through a series of rescaling
and minimization steps the structures are allowed to relax to lower e
nergy configurations with standard bond lengths and reduced van der Wa
als clashes. This algorithm has been tested on the variable loops of y
east tRNA-Asp and yeast tRHA-Phe, as well as the sarcin-ricin tetraloo
p and the anticodon loop of yeast tRNA-Phe. The results indicate good
correlation between potential energy and the loop structure prediction
s that are closest to the variable loop crystal structures, but poorer
correlation for the more isolated stem loops. The number of stacking
interactions has proven to be a good objective measure of the best loo
p predictions. Selecting on the basis of energy and stacking, we obtai
n two structures with 0.65 and 0.75 Angstrom all-atom rms deviations (
RMSD) from the crystal structure for the tRNA-Asp variable loop. The b
est structure prediction for the tRNA-Phe variable loop has an all-ato
m RMSD of 2.2 Angstrom and a backbone RMSD of 1.6 Angstrom, with a sin
gle base responsible for most of the deviation. For the sarcin-ricin l
oop from 28S ribosomal RNA, the predicted structure's all-atom RMSD fr
om the nmr structure is 1.0 Angstrom. We obtain a 1.8 Angstrom RMSD st
ructure for the tRNA-Phe anticodon loop. (C) 1996 John Wiley & Sons, I
nc.