Predicting Nucleic Acid Hybridization and Melting Profiles

Many applications in modern biotechnology require a rapid and sensitive prediction of hybridization or partial hybridization between an oligonucleotide and potential targets in a genomic DNA or mRNA database. In addition, the accurate prediction of melting profiles between an oligonucleotide and a target nucleic acid is also of great value. DNA and RNA chip technologies [16, 17], PCR primer design, sequencing by hybridization and gene diagnostic methods, including SNP detection, are all technologies for which these predictions are very important. DNA chips alone have numerous applications. They are useful to monitor whole genome gene expression [15]. They are well adapted to the detection of single nucleotide polymorphisms (SNPs) [8], to identifying organisms from their sequences [2] or the characterization of splicing variants [5]. They can be used for DNA sequencing [3] or to search for protein target sites on DNA [7]. Computational methods for hybridization and melting prediction tend to make use of existing tools. Thus the very well-known BLAST [1] program is used for database searching to determine oligonu-cleotide specificity. BLAST or MegaBLAST are inappropriate methods since they were designed to search for similarity based on an evolutionary model rather than hybridization based on equilibrium thermodynamics. Similarly, melting temperatures for folded, single stranded oligonucleotides, or hy-bridized pairs of single stranded nucleic acids are usually determined using simple two state models that assume a single, presumably minimum energy folding, or hybridization versus a totally unfolded or non-hybridized state. In the case of hybridization, the target is assumed to be the reverse complement , with possibly a few mismatches allowed, so that no computation is needed to determine the hybridized state. In some cases, melting temperatures are estimated by using ad hoc methods based on GC content. We have undertaken a research program to provide more sophisticated and therfore, we hope, more accurate algorithms and software to address the above issues. In the case of DNA or RNA melting, the work is also of theoretical interest, since we are attempting to determine entire melting profiles, and not just melting temperatures. In addition, we are attempting to determine thermodynamic values, such as the free energy, enthalpy and entropy changes during melting. These can be measured using calorimetric methods. We have already created a prototype version of a program that we have named " FASTH " (FAST Hybridization). The name was deliberately chosen because of similarities to the existing FASTA software [9, 11, …