Analysis of phylogenetic signal in protein language model embeddings
Brendonas Stakauskas, Paweł Górecki · 2026
Read Analysis of phylogenetic signal in protein language model embeddings by Brendonas Stakauskas, Paweł Górecki (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/PMC13369888.txt.
Description
Brendonas Stakauskas and Paweł Górecki test in 2026 if the deep codes that protein learn mods build hold the marks of kin trees. Such mods learn shape and work from code with no kin rules built in. The pair checks if gaps in code space can rebuild trees in a no train pass.
One code lump per protein loses most kin marks, they find: tree shape and limb spans come back poor. Stack gaps by each code spot in line, in mods that read the line whole, lifts shape truth to par with trained tools. Limb spans stay shaky. Kin use of such codes needs the stack route.
Catalog record
| CREATOR | Brendonas Stakauskas, Paweł Górecki |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31481 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |