Multiplex primer prediction software for divergent targets

Gardner, Shea N.; Hiddessen, Amy L.; Williams, Peter L.; Hara, Christine; Wagner, Mark C.; Colston, Bill W.

For runs predicting universal viral primers including all viruses at once in the target set, all viral complete genomes and segments downloaded from publicly available sequence databases (Genbank, Baylor, TIGR) as of 25 April 2007 were used. Draft sequences with multiple contigs were merged into a single sequence entry, with contigs separated by 1000 N's, a stretch sufficiently long (>d 2 ) so that primer pairs would not be designed to fall on different contigs, although there were very few draft sequences in contigs where this was necessary. Because of the large numbers of sequences in two families, only the MP segment sequences from Orthomyxoviridae and the L segment from Bunyaviridae were included, as these are the more conserved segments, reducing the number of targets by 23 017 sequences. The total number of target sequences for these all virus runs were 11 477. We predicted a multiplex set of universal 10-mers without binning the primers into separate reactions, but it was necessary to use a very low x dimer and x homodimer of À11 kcal/mol to be possible to predict multiplexed primers to amplify every target. For the next calculations, we subdivided the primers into sub-reactions with 20 primers per reaction bin, to avoid excluding primers that were predicted to form primer dimers, and raised x dimer and x homodimer to -7 kcal/mol. Primer sets of length 5-18 were predicted ( Figure 1 ). To assess the effect of removing T m constraints, we generated universal primer sets with length but no T m requirements, and compared the primer counts to those with T m constraints ( Figure 2 ). Primer sets are available by contacting the authors.