Two developments from 3 November this week: a regulatory first for an oral GLP-1 medicine in Australia, and a careful look at how well AI can read the amino acid order of a peptide.
Australia clears the first oral GLP-1 tablet
On 3 November 2026, Australia's Therapeutic Goods Administration (TGA) registered oral Wegovy, the semaglutide tablet from Novo Nordisk. ABC News reports it is the first GLP-1 medicine the TGA has approved in oral form. It is cleared for adults with obesity and for people who need to lower cardiovascular risk, and it needs a doctor's prescription.
Four strengths were registered: 1.5 mg, 4 mg, 9 mg and 25 mg. The TGA has yet to review packaging and production processes, which must meet quality standards before the product can be released. No price has been set for Australia, and the report says the tablet is expected on shelves before Christmas.
Australia's Pharmaceutical Benefits Scheme currently subsidises weight-loss drugs only for type 2 diabetes. Eli Lilly's oral weight-loss drug, orforglipron, is still awaiting a TGA decision.
How well can AI read a peptide's sequence?
A paper in Nature Methods, published on 3 November, benchmarked eight deep-learning models for de novo peptide sequencing. This is the task of working out a peptide's amino acid order directly from mass spectrometry data, without looking it up in a database. The comparison used three public datasets.
Most sequencing errors were not specific to any one model. They were driven by missing fragment ions, the pieces that mark each position along the chain when a peptide is broken apart in the instrument. Where those are missing, the sequence stays ambiguous whichever model is used.
Residues that look alike and have equal mass also remain hard to resolve, because mass alone cannot tell them apart. The results suggest that better data, not only better models, is needed.