Summary of conserved local base-pairing interactions in HIV-1 PR and RT wild-type consensus sequences

The MFOLD thermodynamic folding algorithm, which predicts RNA structure based on minimum free energy on folding, was used to detect potential local base pairing interactions between RNA nucleotides in the PR-RT region. Mfold analysis was carried out using a sliding window scanning process (implemented in perl) to minimize any arbitrary effects of the region chosen for analysis. Stepwise scanning in a 5’ to 3’ direction with a ‘sliding window’ of fixed length was used to generate subsequences within the region of the HIV genome under analysis (codons 1-99 of PR and 1-240 of RT) and 500bp of flanking sequence either side (derived from subtype consensus sequences available from the Los Alamos HIV Sequence database). A step size of 20bp was used in combination with a range of window sizes (300, 600, 900 and 1200bp). Base pairing interactions within predicted structures generated using this approach were then analysed. The frequency of base-paring between pairs was calculated across each of the windows in which the pair occurred. When a range of structures was predicted for a sequence under analysis, the average frequency of pairs across each of the predicted structures was scored. Base pairing interactions that were consistently predicted across all windows and structures (>90%) were considered robust. With MFOLD, all sequences were folded at 37 degrees Celsius using default parameters.