An Updated Recursive Algorithm for RNA Secondary Structure Prediction with Improved Thermodynamic Parameters

David H. Mathews¹, Troy C. Andre¹, James Kim¹, Douglas H. Turner^1*, and Michael Zuker²

¹Department of Chemistry, University of Rochester, Rochester, NY 14627-0216

²Institute for Biomedical Computing, Washington University, St. Louis, MO 63110

*Author to whom correspondence should be addressed.

An updated recursive algorithm that minimizes free energy predicts 82.5% of phylogenetically determined base pairs from sequence in four small subunit rRNAs, four group I introns, three group II introns, and 41 tRNAs. The rRNAs and group II introns were folded in phylogenetically determined domains of no more than 500 nucleotides. The algorithm incorporates recently determined thermodynamic parameters for the free energies of internal loops of 2 by 1 and 2 by 2 nucleotides. New free energy bonuses for tetraloops and triloops have been developed by consideration of the database of phylogenetically determined structures. Finally, new rules for coaxial stacking have been applied. This new version will be available in FORTRAN for Unix machines and a C++ version is now available for use on Personal Computers with Windows 95 or Windows NT. The program was used to explore structures predicted to have a free energy near the minimum. On average, a structure with 92% of phylogentically determined base pairs is found within 2% of the minimum free energy. For a roughly 400 nucleotide RNA, this is typically 2.3 kcal/mol above the minimum free energy. Implications for determining RNA secondary structure from sequence are discussed.