Same inference, cheaper walk on similar sequences
The peer-reviewed Nature paper published 16 September 2026 recasts Bayesian phylogenetics as Delphy. The load-bearing part is an explicit mutation-annotated tree that exploits how alike epidemic sequences already are, so the sampler spends less time walking through differences that are not there.[1]
Reruns of Ebola, Zika, SARS-CoV-2, mpox and H5N1 analyses kept state-of-the-art accuracy while finishing 2–3 orders of magnitude faster. A simulated 100,000-sequence set completed within a day. Those are the measured gears. They are not a claim that the next outbreak will be dated in an hour on a laptop.[1]
The method is public; the hour on a laptop is not
RSA-260's factor checked out from a social post while the factoring method stayed unpublished, so the cost curve stayed unreadable. Delphy is the opposite gear: the algorithm, the tree, and the client-side web app that runs on the user's machine are in the paper. The remaining gap is whether independent labs reproduce the same 2–3 order cut on their own alignments.[1], [2]
If an independent group publishes a wall-clock match on a 100,000-sequence epidemic alignment within a day on commodity hardware by 2027-03-31, the simulated day becomes a laboratory hour. If the match needs a cluster, the web-app promise stays a convenience wrapper around the same core.[1]