Our study did not begin with a single headline result in mind. It grew from a broader question: how do human neural progenitor cells become neurons over time, and what happens if this process unfolds in the presence of amyloid-β (Aβ), a peptide strongly associated with Alzheimer’s disease?
Rather than focusing only on end-stage pathology, we wanted to observe the dynamics of early neuronal differentiation. Using human iPSC-derived neural progenitor cells, we followed differentiation across four time points (days 0, 7, 13, and 20). In parallel, we exposed a matched set of cells to Aβ 1–42 from the start of differentiation. This design allowed us to compare not just final states, but the transitions between them.
To capture these changes at high resolution, we generated a multimodal single-cell dataset. Single-cell RNA sequencing (scRNA-seq) allowed us to profile gene expression in tens of thousands of individual cells, revealing structured differentiation trajectories. Single-cell ATAC sequencing (scATAC-seq) provided complementary chromatin accessibility information, offering regulatory context for the transcriptional programs we observed.
One of the most interesting aspects of the project was what we did not see. While Aβ exposure was associated with structured differences in gene expression at specific differentiation steps, we did not detect strong global differences in chromatin accessibility between conditions. Rather than diminishing the value of the epigenomic data, this reframed its role: in this dataset, scATAC-seq primarily serves as a regulatory-prioritization layer, supporting transcription factor and enhancer–gene network analyses rather than highlighting dramatic accessibility shifts.
A significant part of the work happened behind the scenes. Single-cell experiments require careful quality control, harmonized processing, and transparent documentation to ensure that observed patterns reflect biology rather than technical variation. Because of this, we focused on building a resource that others can reuse confidently.
All raw and processed data are publicly available (GEO: GSE307094), along with metadata, clustering annotations, trajectory groupings, and analysis code. By sharing a time-resolved, multimodal single-cell map of human neuronal differentiation under baseline and Aβ-exposed conditions, we hope to provide a foundation not only for studying disease-relevant perturbations, but also for benchmarking computational methods and exploring regulatory mechanisms in early human neurogenesis.