Unify learns cellular evolution with universal multimodal embeddings

Huawen Zhong, Wenkai Han, Guoxin Cui, David Gomez-Cabrero, Jesper Tegner, Xin Gao, Manuel Aranda · 2026

Read Unify learns cellular evolution with universal multimodal embeddings by Huawen Zhong, Wenkai Han, Guoxin Cui, David Gomez-Cabrero, Jesper Tegner, Xin Gao, Manuel Aranda (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/PMC13527131.txt.

Description

Zhong and colleagues introduce Unify in 2026, a machine learning method that compares single-cell gene data across species. Comparing cell atlases between species is hard because genes diverge over evolution and lab batches differ. Unify builds shared cell profiles called macrogenes, which group genes by function rather than by one-to-one matching.

The method blends gene activity data with embeddings, meaning numeric summaries, from protein and general language models. It removed batch effects while keeping real biological signals in species separated by over 700 million years of evolution. Unify also predicted how mouse cells respond to perturbations better across species and rebuilt more accurate cell-type family trees. The authors present it as a tool for evolutionary biology and cross-species genomics.

Catalog record

CREATORHuawen Zhong, Wenkai Han, Guoxin Cui, David Gomez-Cabrero, Jesper Tegner, Xin Gao, Manuel Aranda
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE71 KB
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13527131

Browse the full library · About the Tower of Babel