models.fit_scage_reference

models.fit_scage_reference(
    data,
    *,
    age_col='age',
    min_abs_r=0.2,
    max_sites=30000,
)

Fit the per-CpG age models on a bulk reference cohort.

Parameters

Name Type Description Default
min_abs_r float Keep only sites whose absolute correlation with age clears this. Including every site drowns the signal in slopes that are noise. 0.2
max_sites int Cap, applied after ranking by \|r\|. Thirty thousand informative sites is far more than any single cell will cover, and the likelihood is a sum over the covered ones only. 30000