Phylogenetic inference with RAPDs: Some observations involving computer simulation with viral genomes: Some observations involving computer simulation with viral genomes

The technique of random amplification of polymorphic DNA (RAPD) or RAPD- PCR is currently a popular essay for DNA variation. Although there has been considerable experimental application of this technique for phylogenetic inference, the theoretical and computational verification of this methodology has, so far, been limited to a single study where randomly generated and mutated sequences were used. In order to supplement this study, the performance of RAPD analysis was assessed on complete vital genomic sequences by computer simulation. To this end, a simulated RAPD dataset was generated with 32 primers (4 bp) from seven viral genomes (genome size approximately 11,000 bp) from the Flavivirididae and compared to a random sequence of 10,800 bp. There was a detectable difference in the size distributions and number of RAPD fragments between the viral genomes and the random sequence, but this was not judged to be statistically significant. Therefore, for these flaviviruses, their structured and nonrandom nature has not violated the predictive value of a renewal theory model. The phylogenetic relationships inferred with RAPDs were compared to those inferred from nucleotide sequence alignment for congruence and robustness. Nucleotide divergence estimates inferred from RAPDs and sequence alignment were correlated (r = 0.719); however, unlike sequence data, RAPD data often produced nonadditive distances. A dendrogram generated by the weighted least squares method of Fitch and Margoliash was reasonably congruent with the dendrogram generated from sequence data and only differed in the branching order of the four Dengue viruses. Though this RAPD dendrogram suffered from a reduction of clarity, it appears that RAPD analysis, at least in principle, has recovered s phylogeny.