Correction, 4 October 2026. This article was first published 29 March 2026 claiming to validate Jean-Claude Perez's 2020 numerical observation. The following claims are withdrawn. The "82.6 percent validation" figure was a ratio of two percentage differences (26.1 divided by 31.6), which is not a validation metric. The Z-scores of 68 to 91 were artifacts of comparing viral genes against a housekeeping-gene baseline (22.1±3.4 ruptures/kb) that every viral sequence violates by two orders of magnitude; they are not quotable, and the article's own results table was irreconcilable with that baseline. "All natural viruses use -1 PRF" is false: programmed +1 frameshifting exists in nature, the bacterial release-factor-2 gene prfB being the standard textbook example. The "8-10 percent frameshifting efficiency" figure misstated Mulroney et al. (2024), which reports frameshifted product at about 8 percent of in-frame reporter protein in a reporter construct, not a translation rate; the same correction was applied to two other articles on this site the same day. The gain-of-function framing was wrong: the 2P substitutions are published, intentional prefusion-stabilisation design, and the product is a vaccine, not a virus. A melanoma tumour-growth claim attributed to Rubio-Casillas et al. was removed as unverified against the paper, which is a narrative review. A "Monroe et al. 2024" reference could not be traced and is removed. An "A-" self-grade and a "new scientific contribution" claim are removed. What survives is below, stated at the strength the evidence bears.
...