Candidate Variant Prioritization and Tissue-Mechanics Endpoints for Functional Validation in Scoliosis: A Computational Reanalysis

We reanalysed published human variant records and zebrafish tissue-mechanics data to prioritize substitutions for testing and select readouts for subsequent functional assessment. We curated 121 variant records, scored 102 substitutions and retained 56 for the primary structural comparison. Four substitutions exceeded 1 kcal mol−1 in ThermoMPNN and EvoEF: FLNB L582P/Y1294C and FNDC1 F108S/Y1880C. Targeted DynaMut2 predictions exceeded 1 for Y1294C and F108S, while Y1880C was reference-sensitive. Reanalysis of published sspo, katnb1 and vangl2 fish measurements reproduced increases in mean squared shear-wave velocity and shifts associated with N-acetyl-L-cysteine ethyl ester (NACET) treatment. We evaluated a geometry-based calibration using two untreated group means and two withheld treatment means. In 20 mutant fish measured after curve onset, velocity was associated with the most severe sagittal or coronal Cobb angle (R2 = 0.698). Leave-one-animal-out root-mean-square error was 24.6° (95% internal resampling interval, 17.5°–35.0°), versus 42.3° for the training-mean baseline. The small, post-onset, mixed-plane dataset does not establish prognosis or direction of causation. These findings guide candidate selection and tissue-level testing; they do not test cross-scale association between human substitutions and fish phenotypes.