A critical review of PCR primer design algorithms and cross- hybridization case study

The success of the polymerase chain reaction (PCR) is highly dependent on primer design. Commonly used primer design programs rely upon a core set of parameters such as melting temperature, primer length, GC content, and complementarity to optimize the PCR product, but weight those parameters to differing degrees, as well as include other parameters for PCR specific tasks. An analysis of these design algorithms ,and other available PCR primer analysis software, was conducted to find the best means of predicting non-specific PCR products in a laboratory environment using Saccharomyces cerevisiae deletion strains. Results here show that there is no web-based program that is well-suited to the task of post-design PCR analysis for non-specific annealing and secondary structure in the context of the whole genome. A brute force technique was employed using the National Center for Biotechnology Information’s (NCBI’s) Megablast and Zuker’s mfold to correlate primer sequence similarities and secondary structure predictions in a full genome context, with inconclusive results. The ultimate conclusion is that there is a need for a user-friendly, post-design PCR analysis program to accurately predict nonspecific hybridization events that impair PCR amplification. INTRODUCTION The ability to reproduce a target section of a DNA sequence through the use of the polymerase chain reaction has facilitated a wide array of amplification techniques. Whether the objective is shotgun sequencing, or target specific, the success of the PCR strategy is highly dependent on the small synthetic oligonucleotides that hybridize to the complementary DNA sequence. These short nucleotides function in pairs known as the forward and reverse primers, which amplify a specific DNA sequence and, more importantly, anneal exclusively to that DNA target locus (Lexa, 2001). The primer pairs are designed and selected so that they extend toward each other, polymerizing the complementary DNA sequence to the extent that the target region is covered in each cycle of the PCR. Each cycle begins at a high temperature (~95°C) to denature the double-stranded DNA into two single strands. This is followed by a lower temperature step (45-65°C) in which the primers anneal to their respective complementary DNA sequence. The temperature is then increased (~75°C) to enable the primers to extend by polymerizing nucleotides complemetary to the target DNA as shown in Figure 1. This process is then repeated for 25-45 cycles (Glick and Pasternack, 1998). After recently joining the NASA Functional Genomics Group at the Stanford F. John Burpo A critical review of PCR primer design algorithms __________________________________________________________________________________ 2 Figure 1 (Kampke, et al, 2001) Figure 1 (Kampke, et al) Genome Center, I became involved in a project to clone all of the deletion cassettes in the 6,000(+) Saccharomyces cerevisiae deletion strains in order to then sequence the amplification products and verify the inserted molecular bar codes. A lab colleague had optimized two 96-well test plates with two corresponding sets of primers (each well contained a separate deletion strain matched to a corresponding well in two plates of primers). The optimized process resulted in a number of seemingly non-specific PCR products for many of the wells. I then became interested in analyzing available primer design algorithms and determining the best means to predict and avoid similar non-specific products for the remaining deletion strains. The objective of this paper is to analyze the algorithms and parameter weightings of commonly used primer design programs, such as PRIDE, PRIME+, DOPRIMER, PRIMO, Primer Master, and MEDUSA, and then apply these tools to the primer sets used in the NASA Functional Genomics lab. The longer-range objective is to develop a predictive model program for non-specific hybridization products. REVIEW OF PCR PRIMER DESIGN Designing the optimal primer pair entails a tradeoff of a variety of parameters. Over the course of the last ten years, a set of parameters has evolved to establish the core of many available programs. These include melting temperature, string-based alignment scores for complementarity, primer length, and GC content. Most programs establish real values for these primer criteria and involve trade-offs to find the optimal primer for a particular use (Kampke, et al, 2001). Only the PRIDE program was found to use qualitative user inputs and a fuzzy logic in its algorithm (Haas, Vingron, et al, 1998). Many programs include additional parameter objectives such as minimizing the total number of primers for a project (and therefore cost), excluding various target sections (repeat rich regions, or GC content <20 or >80%), target length, and so forth to improve primer quality (Haas, Vingorn, et al, 1998). Once the parameters and weightings are set by the user, these algorithms iteratively cycle through all primer candidates to identify the optimal primer set with the highest objective score.