It is therefore expected that, of the four tests, Test 3 would be the least likely to produce any evidence of secondary structures being less disrupted in the M-recombinants than in the S-recombinants. It is noteworthy that these tests only indicated significant evidence of lower degrees of folding disruption in the Mrecombinants than would be expected under random BRD7552 recombination when we considered the structure of the complete HIV-1 coding region. When we applied these and related tests to individual genome regions corresponding to known biologically functional structural elements, they yielded no evidence that that M-recombinants had less disrupted structural elements than S-recombinants. While it is possible that, relative to selection favouring maintenance of proper protein folding, selection favouring the maintenance of biologically functional RNA secondary structures has a much smaller influence on patterns of recombination in HIV, it is also possible that our RNA folding disruption tests were simply less powerful than the protein folding disruption tests. In this regard, the RNA folding disruption tests had four potentially important shortcomings: the actual parental sequences of naturally occurring recombinants were not used in these tests and it is entirely possible that with these in hand subtle structural differences between the actual and simulated recombinant genomes would have been clearer; the individual recombination events that were analysed involved single pairs of 59 and 39 recombination breakpoints and were probably not representative of natural recombinants which frequently have more than two detectable breakpoints; the accuracy of computational secondary structure prediction is not perfect and it is likely therefore that incorrectly G3335 inferred base-pairing interactions decayed the power with which disruptions of actual base-pairing interactions could be estimated; the possible inclusion of nonviable viruses within the set of analysed sequences could have decreased the power of our tests because it would have violated the implicit assumption that all of the analysed HIV genomes were reasonably fit and were therefore likely free of recombinationinduced RNA structure disruption.