Pierre Nicodeme and Jean-Marc Steyaert
We investigate the problem of designing efficient multiplex PCR for medical applications. We show that the problem is NP-complete by transformation to the Multiple Choice Matching problem and give an efficient approximation algorithm. We developed this algorithm in a computer program that predicts which genomic regions may be simultaneously amplified by PCR. Practical use of the software shows that the method can treat 250 non-polymorphic loci with less than 5 simultaneous experiments.