Disentangling covariate effects on single-cell-resolved epigenomes with DeepDive

Andreas Fønss Møller, Jesper Grud Skat Madsen · 2026

Read Disentangling covariate effects on single-cell-resolved epigenomes with DeepDive by Andreas Fønss Møller, Jesper Grud Skat Madsen (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/PMC13576652.txt.

Description

Andreas Fønss Møller and Jesper Grud Skat Madsen present DeepDive, a 2026 deep learning framework for untangling which factors shape single-cell chromatin data. Chromatin accessibility maps show which gene switches are open in each cell. In human cohorts, factors such as obesity and diabetes travel together, so standard tools cannot cleanly separate their effects.

DeepDive rebuilds the accessibility maps while splitting variation into named and hidden sources. It beat leading methods even when some factor labels were missing, and recovered true signals from heavily tangled test data. The split representation also allows what-if experiments: swapping one factor while holding the rest fixed. Applied to pancreatic islet cells, it ranked factors tied to a diabetes-linked beta cell subtype and named candidate gene regulators. The authors pitch it for hard cohort questions.

Catalog record

CREATORAndreas Fønss Møller, Jesper Grud Skat Madsen
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41525
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13576652

Browse the full library · About the Tower of Babel