Sequence-Intrinsic Thermodynamic and Mechanical Constraints Resolve Clinically Classified Missense Variants Unresolved by Conservation-Based Predictors

Read full text
Authors: Suhail Bachani
Cited by: 0

Abstract

Background: Accurate interpretation of missense variants remains a central challenge in clinical genomics. Conservation-based ensemble predictors achieve high accuracy overall but systematically fail on 15-25% of variants that fall within indeterminate score ranges (0.4-0.6), representing a substantial fraction of Variants of Uncertain Significance (VUS). We hypothesized that sequence-intrinsic biophysical properties–independent of evolutionary history–could resolve pathogenicity in these conservation-blind cases. Methods: We analyzed 2,000 high-confidence ClinVar missense variants with matched REVEL scores using a deterministic analytical framework grounded in polymer biophysics. The framework quantifies local thermodynamic and mechanical perturbations from seven sequence-intrinsic features: nucleotide transition free energy (Delta-G), GC-content perturbation (thermal stability), codon position impact (translational kinetics), local sequence stiffness (persistence length proxy), harmonic balance (base-stacking symmetry), compositional complexity (Shannon entropy), and neighbor-codon elemental transitions (kinetic coupling). Results: Within the REVEL grey zone (0.4-0.6, n=145), where conservation-based predictors are uninformative by definition, sequence-intrinsic scores distinguished pathogenic from benign variants with T=3.62 (p=5.1x10^-4) and AUC=0.67. Low correlation with REVEL (r=0.27) confirmed orthogonality. Gene-stratified analysis revealed 40% stronger discrimination in structurally fragile genes (T=10.30) versus mechanically robust genes (T=7.33), consistent with biophysical rather than evolutionary constraint mechanisms. Nucleotide transition severity–reflecting base-stacking free energy differences–provided the strongest individual signal (T=9.51). Conclusions: Thermodynamic and mechanical properties encoded at the DNA sequence level represent an independent and biologically meaningful axis of information for variant interpretation, particularly where evolutionary signal is weak or absent. These sequence-intrinsic constraints likely operate through effects on mRNA secondary structure stability, ribosomal translocation kinetics, and co-translational folding efficiency.

Fields of Study & Subjects

Genomics and Rare Diseases Hereditary Neurological Disorders Genetic factors in colorectal cancer

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
3 January 2026
Document type
Preprint