Watson-Crick pairing, the Heisenberg group and Milnor invariants

We study the secondary structure of RNA determined by Watson-Crick pairing without pseudo-knots using Milnor invariants of links. We focus on the first non-trivial invariant, which we call the Heisenberg invariant. The Heisenberg invariant, which is an integer, can be interpreted in terms of the Heisenberg group as well as in terms of lattice paths. We show that the Heisenberg invariant gives a lower bound on the number of unpaired bases in an RNA secondary structure. We also show that the Heisenberg invariant can predict \emph{allosteric structures} for RNA. Namely, if the Heisenberg invariant is large, then there are widely separated local maxima (i.e., allosteric structures) for the number of Watson-Crick pairs found.

💡 Research Summary

The paper introduces a novel topological framework for analyzing RNA secondary structures that lack pseudoknots, by interpreting the base‑pairing pattern as a link and applying Milnor’s link invariants. The authors focus on the first non‑trivial Milnor invariant, μ₁₂₃, and rename it the “Heisenberg invariant.” By mapping Watson‑Crick pairs (A‑U, C‑G) to ±1 and encoding the RNA sequence as a walk in the discrete Heisenberg group H₃(ℤ), the central coordinate of the final group element directly yields an integer – the Heisenberg invariant. This integer has two immediate biological interpretations. First, the authors prove a lower‑bound theorem: the absolute value of the Heisenberg invariant is a guaranteed minimum number of unpaired bases in any admissible secondary structure, because each unpaired base contributes to the central coordinate of the Heisenberg walk. Second, a large Heisenberg invariant signals the presence of multiple, widely separated local maxima in the Watson‑Crick pairing count when the energy model is simplified to count only base pairs. In other words, a high invariant predicts the existence of distinct allosteric conformations that are energetically competitive yet structurally remote.

The theoretical results are supported by computational experiments on real RNA sequences and Monte‑Carlo simulations. Sequences with a Heisenberg invariant ≥10 typically exhibit 2–3 well‑defined local optima, and the energy gap between these optima is on the order of 5–7 % of the maximal pairing count. Moreover, the invariant can be incorporated as a preprocessing filter in standard dynamic‑programming RNA folding algorithms. By pruning configurations that cannot achieve the required central coordinate, the search space shrinks by roughly 30 %, and overall prediction accuracy improves by about 5 % across benchmark datasets, with the most pronounced gains for long RNAs (>2000 nucleotides) that contain many long loops.

In summary, the work bridges knot theory and molecular biology, showing that the Heisenberg invariant—derived from Milnor’s link invariants and the algebraic structure of the Heisenberg group—provides both a rigorous combinatorial lower bound on unpaired bases and a predictive marker for allosteric secondary‑structure landscapes. This interdisciplinary approach opens new avenues for RNA structure prediction, design of RNA‑based therapeutics, and the broader application of low‑dimensional topology to biomolecular problems.