Reconstructing developmental and disease progression with sample-level embeddings

Single-cell genomics is transformative for characterizing cellular heterogeneity, but many biological questions require comparing samples rather than individual cells. Typical analyses compare "case" and "control" groups, ignoring sample-level variation within them. Here, we present scSLIDE (single-cell sample-level inference using density estimation), a framework that transforms each sample's single-cell data into a compact profile describing where its cells are distributed in high-dimensional space. By comparing these density profiles, scSLIDE calculates embeddings to characterize variation across samples. We implement scSLIDE in version 6 of our Seurat package and apply it to COVID-19 infection, Alzheimer's disease, and zebrafish embryogenesis. scSLIDE reveals patient subgroups, reconstructs disease trajectories, and identifies coordinated cellular programs across samples. We discover separable axes of infection and severity, a trajectory of Alzheimer's progression, and embryonic "pseudostages" that vary across and within time points. Our results demonstrate how case-control analyses collapse rich biological variation into binary labels and how sample-level embedding represents a powerful analytical alternative.