ORCID: [Your ORCID ID] ABSTRACT Background: The genetic code comprises 22 amino acids (20 canonical plus Selenocysteine and Pyrrolysine). We hypothesized that arranging these amino acids in a circular wheel and classifying positions as "Prime" (positions 2, 3, 5, 7, 11, 13, 17, 19) or "Forbidden" (all others) would reveal non-random patterns in cancer mutation distribution. Methods: We analyzed 190,432 missense mutations from the COSMIC database (v103) across 17 major cancer driver genes. For each mutation, we determined whether the wild-type and mutant amino acids occupied the same category (Prime-to-Prime or Forbidden-to-Forbidden) or crossed categories (Prime-to-Forbidden or Forbidden-to-Prime). We performed 1,000 random simulations for comparison. Results: Of 190,432 mutations, 131,824 (69.2%) preserved their wheel category, compared to an expected 53.7% by random chance (Z-score = 136.0, p < 10⁻⁴⁰⁰⁰). Zero of 1,000 random simulations matched the observed pattern. We observed profound asymmetry: Forbidden-to-Prime mutations (19,389) were 2.0-fold rarer than Prime-to-Forbidden mutations (39,219; χ² = 6,709.5, p < 0.001). The rare F→P mutations include the most clinically significant mutations in cancer biology (IDH1 R132H, TP53 R175H, EGFR T790M). Conclusions: Cancer driver mutations exhibit highly non-random distribution on the 22-position amino acid wheel. Same-category preservation suggests fundamental constraints on viable amino acid substitutions. The asymmetric rarity of F→P mutations indicates strong selective pressure against mutations into Prime positions, with exceptions representing the most oncogenically powerful mutations. This framework may complement existing variant pathogenicity prediction tools. Keywords: cancer mutations, COSMIC database, amino acid wheel, variant pathogenicity, computational biology, genetic code