This paper is an analysis of a model of the regeneration of damaged DNA during the polymerase chain reaction (PCR) made to estimate the degree to which recombination between similar but not identical sequences ("jumping PCR") compromises the results obtained with the technique. The analysis is applicable to museum or field-weathered specimens and also to the damage, such as single-strand nicks, which can occur during storage. Two main results are derived: (i) oligonucleotide priming competes extremely effectively against regeneration derived from priming by overlaps of damaged DNA strands. (ii) Extending result (i), it is shown that jumping PCR is not likely to cause inaccurate sequence determination in standard PCR protocols. One exception arises where it is required to determine the sequence of one specified variant of a piece of DNA which is present in many thousand copies. But, generally, experiments can be performed with confidence that the results are very rarely distorted by jumping PCR.
The basic model is extended to investigate how to increase the average length of DNA fragments input to PCR experiments. Increased input of less-damaged material to the PCR can theoretically be achieved by performing several cycles of denaturation, re-association and replication on the sample DNA in the absence of exogenous primers. Experimental work will be necessary to confirm this. However, jumping PCR may occur at substantial frequencies during such pretreatment. This problem cannot be avoided but can be reduced in pretreatment protocols by studying sequences (such as mtDNA or rDNA) where multiple copies are subjected to evolutionary homogenizing processes.