The Pan-AD Atlas is a large-scale single-nucleus RNA sequencing resource built from over six million nuclei across 13 human cortical datasets, encompassing Alzheimer’s disease, mild cognitive impairment, and control samples. Using advanced computational pipelines, the atlas integrates diverse datasets into a unified, cell-type annotated reference and applies deep learning–based subtyping to define four molecular subtypes of AD. It enables in-depth characterization of disease biology through differential gene expression, sex-specific analysis, co-expression network mapping, cell–cell communication modeling, and regulatory network inference, offering a comprehensive framework to study Alzheimer’s heterogeneity and pathogenesis.

View UMAP by

  • Broad cell-type (7 cell-types)
  • Fine cell-types (subtypes)
  • Age
  • Sex
  • Study
  • Disease (Control, MCI, AD)
  • Disease-subtype (Control, MCI, AD1-4)
  • Braak stage

Default UMAP visualization

Objectives

Build a comprehensive single-cell atlas to uncover cellular and molecular subtypes of Alzheimer’s disease.

Cohorts & Tissues

Human cortical samples from Alzheimer’s disease, mild cognitive impairment, and control cohorts.

Methods (high level)

Integrated multi-cohort snRNA-seq analysis with deep learning–based subtyping and systems-level characterization.

Data & Reproducibility

Publicly available, rigorously quality-controlled datasets with transparent, reproducible analysis pipelines.

Overview of methods used in the Pan-AD Atlas