Predicting the three-dimensional folding of transfer RNA with a computer modeling protocol.

We have developed a computer modeling protocol that can be used to predict the three-dimensional folding of a ribonucleic acid on the basis of limited amounts of secondary and tertiary data. This protocol extends the use of distance geometry beyond the domain of NMR data in which it is usually applied. The use of this algorithm to fold the molecule eliminates operator subjectivity and reproducibly predicts the overall dimensions and shape of the transfer RNA molecule. By use of a replacement pseudoatom set based on helical substructures, a series of transfer RNA foldings have been completed that utilize only the primary structure, the phylogenetically deduced secondary structure, and five long-range interactions that were determined without reference to the crystal structure. In a control set of foldings, all the interactions suspected to exist in 1969 have been included. In all cases, the modeling process consistently predicts the global arrangement of the helical domains and to a lesser extent the general path of the backbone of transfer RNA.