Answers are grounded in this paper and cite their sources.
A protein's sequence says what a gene could do in another species; the cells of that species say what it does there. Comparative genomics measures the first. Chimera measures the second, a gene's deployment, on the same genes and reports the difference. It aligns single-cell atlases across species in one generative model and returns, for every one-to-one ortholog, the evidence in the counts for a shared expression pattern, and Δɑ, the position of that evidence above or below what the protein's identity predicts. This difference is not sequence constraint under another name: across eight tissues, protein identity tracks dN/dS, whereas Δɑ tracks neither. A gene therefore has two coordinates, not one, and the second can contradict the first: spermatocyte genes such as ZPBP2 keep their stage pattern on less conserved proteins, in eight held-out species, whereas conserved hepatocyte enzymes such as ALAS1 share their pattern but not their level. More broadly, genes of specialized cells sit above what their proteins predict in 23 of 24 cohorts, and approved drug targets do so more often than matched genes. Even among nearly identical proteins, where sequence can no longer tell genes apart, the count evidence still does. Nor does this rest on the model alone: independent data support the count evidence in liver and testis and, in a pre-specified test, Δɑ in liver. With both coordinates, one can ask whether another species uses a gene as humans do, the premise of every animal model of a disease gene or drug target.
No takes yet. Share an insight, caveat, or question.
İnecik et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: