RNA Folding Algorithms Research Articles

The massively parallel genetic algorithm for RNA folding: MIMD implementation and population variation.

A massively parallel Genetic Algorithm (GA) has been applied to RNA sequence folding on three different computer architectures. The GA, an evolution-like algorithm that is applied to a large population of RNA structures based on a pool of helical stems derived from an RNA sequence, evolves this population in parallel. The algorithm was originally designed and developed for a 16384 processor SIMD (Single Instruction Multiple Data) MasPar MP-2. More recently it has been adapted to a 64 processor MIMD (Multiple Instruction Multiple Data) SGI ORIGIN 2000, and a 512 processor MIMD CRAY T3E. The MIMD version of the algorithm raises issues concerning RNA structure data-layout and processor communication. In addition, the effects of population variation on the predicted results are discussed. Also presented are the scaling properties of the algorithm from the perspective of the number of physical processors utilized and the number of virtual processors (RNA structures) operated upon.

Distinct RNA structural domains cooperate to maintain a specific cleavage site in the 3′-UTR of IGF-II mRNAs

The insulin-like growth factor II mRNAs are targets for site-specific endonucleolytic cleavage in the 3′-UTR, which results in a very stable 3′ cleavage product of 1.8 kb, consisting of 3′-UTR sequences and a poly(A) tail. The 5′ cleavage product contains the coding region and is rapidly degraded. Thus, cleavage is thought to provide an additional way to control IGF-II protein synthesis. We had established that cleavage requires two widely separated sequence elements (I and II) in the 3′-UTR that form a stable duplex of 83 nucleotides. The cleavage-site itself is located in an internal loop preceded by two stable stem-loop structures. Furthermore, in a study which was based on RNA folding algorithms, we have shown that there are specific sequence and structural requirements for the cleavage reaction. Here, the functions of the different structural domains in cleavage were assessed by deletion/mutational analyses, and biochemical structure probing assays were performed to characterize better the RNA structures formed and to verify the computer folding predictions. The data suggest that the stem-loop domain contributes to maintain a highly specific c leavage-site by preventing the formation of alternative structures in the cleavage-site domain. Involvement of the nucleotides in the cleavage-site loop itself in non-Watson-Crick interactions may be important for providing a specific recognition surface for an endoribonuclease activity.

Dynamic competition between alternative structures in viroid RNAs simulated by an RNA folding algorithm

The folding pathways of viroid RNAs were studied using computer simulations by the genetic algorithm for RNA folding. The folding simulations were performed for PSTVd RNAs of both polarities, using the wild-type sequence and some previously known mutants with suggested changes in the stable or metastable structures. It is shown that metastable multihairpin foldings in the minus strand replicative intermediates are established due to the specific folding pathway that ensures the absence of the most stable rod-like structure. Simulations of the PSTVd minus strand folding during transcription reveal a metastable hairpin, formed in the left terminal domain region of the PSTVd. Despite high sequence variability, this hairpin is conserved in all known large viroids of both subgroups of PSTVd type, and is presumably necessary to guide the folding of the HPII hairpin which is functional in the minus strand. The folding simulations are able to demonstrate the changes in the balance between metastable and stable structures in mutant PSTVd RNAs. The stable rod-like structure of the circular viroid (+) RNA is also folded via a dynamic folding pathway. Furthermore, the simulations show that intermediate steps in the forced evolution of a shortened PSTVd replicon may be reconstructed by a mechanistic model of different folding pathway requirements in plus- and minus-strand RNAs. Thus the formation of viroid RNA structure strongly depends on dynamics of competition between alternative RNA structures. This also suggests that the replication efficiency of viroid sequences may be estimated by a simulation of the folding process.

Predicting RNA H-type pseudoknots with the massively parallel genetic algorithm.

Using the genetic algorithm (GA) for RNA folding on a massively parallel supercomputer, MasPar MP-2 with 16,384 processors, we successfully predicted the existence of H-type pseudoknots in several sequences. The GA is applied to folding the tRNA-like 3' end of turnip yellow mosaic virus (TYMV) RNA sequence with 82 nucleotides, the 3' UTRs of satellite tobacco necrosis virus (STNV)-2 RNA sequence with 619 nucleotides and STNV-I RNA sequence with 622 nucleotides, and the bacteriophage T2, T4 and T6 gene 32 mRNA sequences with 946, 1340 and 946 nucleotides, respectively. The GA's results match the phylogenetically supported tertiary structures of these sequences.

An annealing mutation operator in the genetic algorithms for RNA folding.

An annealing mutation operator in the genetic algorithms (GA) for RNA folding on a MasPar MP-2 has been designed. The mutation probability descends along a hyperbola with respect to the size of the secondary structure, hence the total number of mutations at each generation drops linearly. Especially for long sequences with thousands of nucleotides as opposed to hundreds of nucleotides, the new mutation operator can make the distribution of free energies over all processors on MasPar MP-2 converge only after hundreds of generations. Based upon this new mutation operator, a technique to terminate the GA is also developed. The new mutation operator runs very efficiently. Some variations of the annealing mutation operator are also discussed.

Rat skeletal muscle selenoprotein W: cDNA clone and mRNA modulation by dietary selenium.

Rat skeletal muscle selenoprotein W cDNA was isolated and sequenced. The isolation strategy involved design of degenerate PCR primers from reverse translation of a partial peptide sequence. A reverse transcription-coupled PCR product from rat muscle mRNA was used to screen a muscle cDNA library prepared from selenium-supplemented rats. The cDNA sequence confirmed the known protein primary sequence, including a selenocysteine residue encoded by TGA, and identified residues needed to complete the protein sequence. RNA folding algorithms predict a stem-loop structure in the 3' untranslated region of the selenoprotein W mRNA that resembles selenocysteine insertion sequence (SE-CIS) elements identified in other selenocysteine coding cDNAs. Dietary regulation of selenoprotein W mRNA was examined in rat muscle. Dietary selenium at 0.1 ppm as selenite increased muscle mRNA 4-fold relative to a selenium-deficient diet. Higher dietary selenium produced no further increase in mRNA levels.

A periodic table of symmetric tandem mismatches in RNA.

The stabilities and structures of a series of RNA octamers containing symmetric tandem mismatches were studied by UV melting and imino proton NMR. The free energy increments for tandem mismatch formation are found to depend upon both mismatch sequence and adjacent base pairs. The observed sequence dependence of tandem mismatch stability is UGGU > GUUG > GAAG > or = AGGA > UUUU > CAAC > or = CUUC approximately UCCU approximately CCCC approximately ACCA approximately AAAA, and the closing base pair dependence is 5'G3'C > 5'C3'G > 5'U3'A approximately 5'A3'U. These results differ from expectations based on models used in RNA folding algorithms and from the sequence dependence observed for folding of RNA hairpins. Imino proton NMR results indicate the sequence dependence is partially due to hydrogen bonding within mismatches.

SSCP primer design based on single-strand DNA structure predicted by a DNA folding program.

To predict alterations in single-strand DNA mobility in non-denaturing electrophoretic gels, Zuker's RNA folding program was modified. Energy files utilized by the LRNA RNA folding algorithm were modified to emulate folding of single-strand DNA. Energy files were modified to disallow G-T base pairing. Stacking energies were corrected for DNA thermodynamics. Constraints on loop nucleotide sequences were removed. The LRNA RNA folding algorithm using the DNA fold energy files was applied to predict folding of PCR generated single-strand DNA molecules from polymorphic human ALDH2 and TPH alleles. The DNA-Fold version 1.0 program was used to design primers to create and abolish SSCP mobility shifts. Primers were made that add a 5' tag sequence or alter complementarity to an internal sequence. Differences in DNA secondary structure were assessed by SSCP analysis and compared to single-strand DNA secondary structure predictions. Results demonstrate that alterations in single-strand DNA conformation may be predicted using DNA-Fold 1.0.

Application of computational technologies to ribozyme biotechnology products

Ribozymes are RNA molecules that act enzymatically to cleave other RNA molecules. The cleavage reaction requires the binding of ribozyme to specific sites on the target RNA through (mostly) Watson-Crick base-pairing interactions. Association of ribozyme with target completes a three-dimensional ribozyme/target complex which results in cleavage of the target RNA. We are employing both computational and experimental approaches to identify sites on target RNA molecules that are open to ribozyme attack and to determine which ribozymes are most active against those sites. Two types of computational technologies are available for aiding in the identification of target sites and design of active ribozymes. First, DNA/RNA sequence analysis software is employed to identify sequence motifs necessary for ribozyme cleavage and to look for sequence conservation between different sources of the target organism so that ribozymes with the broadest possible target range can be designed. Second, RNA folding algorithms are employed to predict the secondary structure of both ribozyme and target RNA in an attempt to identify combinations of ribozyme and target site that will successfully associate prior to ribozyme cleavage. The RNA folding algorithms utilize a set of thermodynamic parameters obtained from measurements on short RNA duplexes; while these rules give reasonable predictions of secondary structure for a small set of highly structured RNAs, they remain largely untested for predicting the structure of messenger RNAs. This paper outlines the current status of designing ribozymes that fold correctly and of locating target sites that are sufficiently unfolded to allow ribozyme cleavage.

Application of computational technologies to ribozyme biotechnology products

Ribozymes are RNA molecules that act enzymatically to cleave other RNA molecules. The cleavage reaction requires the binding of ribozyme to specific sites on the target RNA through (mostly) Watson—Crick base-pairing interactions. Association of ribozyme with target completes a three-dimensional ribozyme/target complex which results in cleavage of the target RNA. We are employing both computational and experimental approaches to identify sites on target RNA molecules that are open to ribozyme attack and to determine which ribozymes are most active against those sites. Two types of computational technologies are available for aiding in the identification of target sites and design of active ribozymes. First, DNA RNA sequence analysis software is employed to identify sequence motifs necessary for ribozyme cleavage and to look for sequence conservation between different sources of the target organism so that ribozymes with the broadest possible target range can be designed. Second. RNA folding algorithms are employed to predict the secondary structure of both ribozyme and target RNA in an attempt to identify combinations of ribozyme and target site that will successfully associate prior to ribozyme cleavage. The RNA folding algorithms utilize a set of thermodynamic parameters obtained from measurements on short RNA duplexes; while these rules give reasonable predictions of secondary structure for a small set of highly structured RNAs, they remain largely untested for predicting the structure of messenger RNAs. This paper outlines the current status of designing ribozymes that fold correctly and of locating target sites that are sufficiently unfolded to allow ribozyme cleavage.

A comparison of optimal and suboptimal RNA secondary structures predicted by free energy minimization with structures determined by phylogenetic comparison

This article describes the latest version of an RNA folding algorithm that predicts both optimal and suboptimal solutions based on free energy minimization. A number of RNA's with known structures deduced from comparative sequence analysis are folded to test program performance. The group of solutions obtained for each molecule is analysed to determine how many of the known helixes occur in the optimal solution and in the best suboptimal solution. In most cases, a structure about 80% correct is found with a free energy within 2% of the predicted lowest free energy structure.

Speeding up the dynamic algorithm for planar RNA folding

The simplest dynamic algorithm for planar RNA folding searches for the maximum number of base pairs. The algorithm uses O( n 3) steps. The more general case, where different weights (energies) are assigned to stacked base pairs and to the various types of single-stranded region topologies, requires a considerably longer computation time because of the partial backtracking involved. Limiting the loop size reduces the running time back to O( n 3). Reduction in the number of steps in the calculations of the various RNA topologies has recently been suggested, thereby improving the time behavior. Here we show how a “jumping” procedure can be used to speed up the computation, not only for the maximal number of base pairs algorithm, but for the minimal energy algorithm as well.

Use of an RNA folding algorithm to choose regions for amplification by the polymerase chain reaction

We have used the program FOLD, which employs the Zuker folding algorithm, to identify regions of stable secondary structure in three chicken proto-oncogene mRNAs: c-src, c-myc, and c-fos. We have found that use of reverse transcriptase to synthesize a cDNA template for amplification by the polymerase chain reaction is successful only if the region chosen for amplification does not contain stem structures with calculated free energies less than −14 kcal/mol.

Some simple computational methods to improve the folding of large RNAs.

Computational methods are described which increase the efficiency of the RNA folding algorithm described by Zuker and Stiegler. Bit addressing has been used to reduce the memory requirements from 2NxN to NxN/2. The order in which the nucleotide sequence is examined internally has been altered, and some additional short arrays which carry temporary information have been introduced. These changes optimize the management of the large data arrays generated by the algorithm. The methods were developed for use with a UNIVAC 1100/82 computer. They are, however, easily adaptable to other computers; especially those with virtual memory capabilities. The analysis of sequences up to 1000 nucleotides long are relatively routine, and larger searches are also feasible. Some limitations and applications of the algorithm are also discussed.