PRIME: An evaluation framework for protein representation inference and generalization in viral mutation space
Kaetlyn Gibson, Po-E Li, Valerie Li, Martha Dix, Li-Wei Hung, George Widgery Stelle, Michal Babinski, Patrick Chain, Bin Hu · 2026
Read PRIME: An evaluation framework for protein representation inference and generalization in viral mutation space by Kaetlyn Gibson, Po-E Li, Valerie Li, Martha Dix, Li-Wei Hung, George Widgery Stelle, Michal Babinski, Patrick Chain, Bin Hu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13425921.txt.
Description
Kaetlyn Gibson and colleagues build PRIME, a fairer test bench for protein language models that predict how viral mutations behave. Standard random splits leak near-identical sequences across train and test sets, flattering scores that then fail in real surveillance.
Using 347,432 sequences from the SARS-CoV-2 spike binding region, the team shows random splits can top a score of 0.90 while failing on fresh mutation sites. Position-stratified checks plus targeted tuning of a 600-million-parameter model are put forward instead. Virus watchers get a stricter yardstick for tools that must spot the next risky variant.
Catalog record
| CREATOR | Kaetlyn Gibson, Po-E Li, Valerie Li, Martha Dix, Li-Wei Hung, George Widgery Stelle, Michal Babinski, Patrick Chain, Bin Hu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |