We have developed a real-time PCR primer design algorithm based on the general guidelines described in Section 1 (Wang and Seed, 2003a). An outline of the algorithm is presented in Figure 5.1. The algorithm is implemented by a design tool called uPrimer, with which we have designed more than 300,000 primers encompassing most human and mouse genes. The following comprise detailed descriptions of the properties of these primers.
Although primer specificity is one of the most important requirements in real-time PCR, most primer design programs take only one target sequence without considering mispriming to off-target templates. At the present time, to address the mispriming problem, one can design a number of primer pairs and then individually check cross-matches of each primer with BLAST (Altschul et al., 1990). However this screening step is incomplete and does not consider some important design criteria. For example, crossmatches at the 3' end of a primer are more likely to produce non-specific amplicons. In our experience, only about two thirds of the primers designed in the conventional way can be used in real-time PCR experiments.
Mismatches are known to significantly affect binding stability, and sometimes even a single mismatch may destabilize a DNA duplex. Therefore contiguous base pairing is expected to be one of the most important contributing factors for duplex stability. Our primary filter for primer cross-reactivity is to reject a stretch of contiguous residues that is also found in off-target sequences. Every possible 15-mer in a primer sequence is compared to all known gene sequences in a genome. Primers with repetitive 15-mers are quickly identified and rejected using a computational hashing technique (Wang and Seed, 2003b). To further reduce cross-reactivity, BLAST searches for primer sequence similarity are carried out and all qualified primers have BLAST scores of less than 30.
The primer 3' end is essential to prevent non-specific amplification because Taq polymerase extension can be greatly retarded by terminal mismatches (Huang et al., 1992). Therefore a more stringent filter is applied to cross-matches at the 3' ends. In our algorithm the cross-hybridizing Tm for the 3' end perfect match does not exceed 46 degrees.
Random priming in RT reactions leads to significant amount of non-coding cDNA products, which are a potential source for mispriming. Because of their abundance, more stringent primer specificity filters are applied against non-coding RNA sequences.
Was this article helpful?