A General Computational Framework for Sequence-Level Thermodynamic Analysis of Modeled RNA-DNA and DNA-DNA Duplex States: A Proof-of-Concept Study in HIV-1

RNA-DNA hybridization is relevant to several molecular processes, but computational energy estimates alone cannot establish hybrid formation in cells. This study presents a computational framework for comparing model-derived RNA-DNA and DNA-DNA energy metrics across 16 selected regions of the HIV-1 reference genome HXB2 (GenBank accession K03455.1). The operational metric was defined as the difference between the RNA-DNA calculation and the DNA-DNA reference calculation (ΔΔG_model). Across the 16 regions, 15 yielded negative metric values and one yielded a positive value. The mean was −8.00 kcal/mol (sample standard deviation, 5.45 kcal/mol). A two-sided one-sample t-test against zero gave t(15) = −5.87, p = 3.08 × 10⁻⁵. The mean remained negative in sensitivity analyses excluding either or both of the overlapping TAR and R5LTR regions. These findings describe a consistent direction in the selected computational panel. However, the regions are drawn from a single reference genome, are not independent biological replicates, and may not represent independent sequence information. Furthermore, our independent verification confirmed that while the RNA-DNA energies are highly reproducible, the DNA-DNA reference energy is highly dependent on the specific algorithmic approach. Eleven of the 16 DNA-DNA reference values showed numerical agreement with standard DuplexFold calculations, while five regions showed discrepancies (1.1–5.3 kcal/mol) that remain unresolved. The results do not demonstrate intracellular RNA-DNA hybrid formation or genomic R-loop formation. Instead, they provide a testable computational hypothesis for further evaluation through reproducible energy calculations and independent experimental measurements.