This paper extends the SEA-AD atlas across ten neo- and allocortical brain regions to determine whether the cellular changes associated with Alzheimer’s disease are shared across the brain. By analyzing the molecular profiles of roughly seven million cells and relating them to regional pathology, the researchers found that vulnerability is remarkably consistent across regions, with the same subsets of inhibitory neurons, oligodendrocytes, and glia changing in a stereotyped sequence that begins before symptoms emerge. The study also uncovered unexpected vulnerability in specialized cell populations, including primary visual cortex layer 4 excitatory neurons previously thought to be resilient, and identified neuronal hyperexcitability as a potential common mechanism linking both early- and late-vulnerable cell types.

This paper describes the creation of a massive molecular and cellular atlas of middle temporal gyrus to better frame and understand how Alzheimer’s disease progresses. By analyzing the genetic activity of millions of individual cells, the researchers discovered that the disease actually unfolds in two distinct phases: an early, slow-moving stage where specific Sst+ interneurons are lost and inflammation begins, followed by a later stage of accelerated neuropathology and loss of excitatory neurons and Pvalb+ and Vip+ inhibitory neuron subtypes. These findings were replicated in prefrontal cortex in SEA-AD and other major AD studies.

This paper represents the characterization of the Alzheimer's pathological spectrum within the caudate nucleus. Utilizing digital pathology, researchers were able to confirm a less pronounced phosphorylated tau and amyloid-β burden when compared to other cortical areas. Phosphorylated tau and amyloid-β were also shown to localize separately into white and gray matter, respectively. Both, in part, explain the lack of neuronal cell death observed in the caudate even in high burden parts of the spectrum. However, researchers were able to identify glimmers of a global phosphorylated tau response in microglia.

This paper introduces B-BIND, a new mathematical framework designed to track how Alzheimer’s disease progresses by analyzing the buildup of various pathological proteins. Because researchers often only have "snapshots" of the brain from different individuals, this model uses a "pseudotime" scale to rank patients along a continuous timeline of disease severity. Ultimately, this framework lays the groundwork for identifying the hidden biological stages of neurodegeneration and the specific cellular changes that drive the disease over time.

This paper introduces TabVI, a lightweight AI model designed to analyze single-cell genomic data more effectively than standard language-based AI. By adapting "transformer" technology—the same tech behind ChatGPT—to better fit the non-sequential, hierarchical nature of genes, TabVI can more accurately identify cell types and biological patterns. This tool provides a more efficient and interpretable way for researchers to study how diseases like Alzheimer’s affect individual cells across large datasets.

This paper introduces scVIP, an AI framework designed to reconstruct how Alzheimer’s disease progresses by combining molecular measurements from individual cells with information about each donor’s disease severity. By learning personalized disease trajectories, scVIP can more accurately predict neuropathology, identify the cell types most affected by disease, and connect findings across independent studies. This approach provides researchers with a powerful new tool for exploring the cellular and molecular changes that drive neurodegeneration.

This paper presents the SEA-AD DREAM Challenge, an open international competition in which 17 teams from 15 countries built models to predict Alzheimer's disease neuropathology from single-nucleus RNA-sequencing data. The best models predicted categorical disease staging with near-perfect accuracy and estimated amyloid-β and tau burden competitively, and the challenge also debuted the first AI Agent Track, establishing a reproducible benchmark for molecular neuropathology prediction.

This paper introduces Annotation Comparison Explorer (ACE), a web-based tool developed by the Allen Institute to address the challenge of comparing cell-type classifications across different brain studies. ACE allows researchers to map their own single-cell data to established taxonomies, such as the Seattle Alzheimer’s Disease Brain Cell Atlas (SEA-AD), and compare annotations like donor demographics and disease-related changes. By applying ACE to multiple published Alzheimer's datasets, the authors identified consistent signatures of the disease, such as a decrease in specific somatostatin interneurons, demonstrating the tool's utility in unifying diverse datasets to advance the understanding of brain health and neurodegeneration.

This commentary on Gabitto, Travaglini, et al 2024 presents SEA-AD as a multifaceted open community resource designed to identify cellular and molecular pathologies that underlie Alzheimer’s disease. Integrating neuropathology, single cell and spatial genomics, and longitudinal clinical metadata, SEA-AD is a unique resource for studying the pathogenesis of Alzheimer’s and related dementias.