PathoEvo turns scattered single-cell disease data into a stage-aware resource
Disease progression rarely moves in a straight line. Tissue microenvironments shift, cell populations expand or contract, and transcriptional programs change long before pathology becomes fixed in gross anatomy. PathoEvo is built around that biological reality, linking single-cell and single-nucleus transcriptomic data to the course of disease rather than treating each dataset as an isolated snapshot.
The new resource brings together 7,452,971 cells or nuclei from 981 samples and 59 datasets, spanning 18 diseases and 27 progression frameworks. Its core purpose is straightforward but consequential: to make heterogeneous single-cell disease studies comparable when they were originally generated under different clinical, pathological, and molecular staging systems. For researchers studying evolution of disease at cellular resolution, that kind of harmonization is not a convenience. It is the difference between isolated observations and a usable map.
A database built to reconcile incompatible staging systems
Single-cell studies have proliferated quickly, but the field has carried a persistent problem. Public datasets often live in separate publications, each annotated with its own staging language, its own pathology conventions, and its own molecular labels. Even when two studies examine the same disease, the stage definitions may not align well enough to support direct comparison. PathoEvo addresses that fragmentation through manual curation, unified processing, metadata harmonization, cell-type annotation, and stage mapping.
The result is a standardized framework centered on disease progression. Instead of forcing users to navigate a patchwork of dataset-specific conventions, the database organizes information around progression frameworks that can be compared across studies. That approach matters because progression biology is often driven by coordinated shifts rather than one dominant lesion. Immune infiltration, stromal remodeling, epithelial state changes, and transcriptional reprogramming can all occur in parallel, and each may look different at distinct stages of the same disease.
One sentence captures the practical value here: disease stage is now searchable, not merely inferable.
Three analytical layers expose cellular, sample, and gene-level change
PathoEvo is not just a catalog of studies. Its data and analytical functions are arranged across three complementary levels that let users move from broad composition to fine-grained molecular behavior. At the cellular level, researchers can examine changes in cellular composition associated with disease stage and cell states. At the sample level, they can characterize molecular variation across progression. At the gene level, they can investigate expression dynamics across relevant cell populations and samples.
This structure is well suited to the way disease evolves in tissue. A shift in the proportion of one cell type can be just as informative as a change in expression of a single pathway gene, especially when those changes occur in a stage-specific pattern. In inflammatory disease, for example, a rising abundance of activated immune cells may track with tissue damage. In cancer, the same logic can apply to transitions in tumor, stromal, and immune compartments, where progression often reflects a moving ecological balance rather than a single molecular event.
PathoEvo is designed to capture that complexity without flattening it. By combining curated metadata with transcriptomic readouts, the platform allows investigators to ask how stage-dependent biology is distributed across cell populations, not only within one study but across many.
Why single-cell progression resources matter for pathology research
Single-cell and single-nucleus RNA sequencing have made it possible to study disease at cellular resolution, but the abundance of data has created its own bottleneck. Without a shared structure for stage assignment and metadata interpretation, comparing datasets becomes laborious and sometimes unreliable. A stage-aware database helps shift the field from isolated exploratory work toward reproducible cross-dataset analysis.
That shift has direct implications for pathological evolution research. If the same disease is represented in multiple cohorts, each with different nomenclature for stage or severity, investigators can miss recurring biological signals or overestimate differences that arise from annotation style rather than biology. A harmonized system reduces that noise. It also gives computational methods a cleaner substrate for downstream analysis, whether the question concerns cellular composition, molecular trajectories, or cell-state transitions.
The database also reflects an increasingly important reality in precision medicine research: progression is often more informative than diagnosis alone. Two patients may share a disease label while occupying different biological stages, and those stages can carry distinct cellular signatures. A resource organized around progression frameworks is therefore better aligned with the biology clinicians and translational scientists actually need to interpret.
Standardization may be the most important feature here
PathoEvo’s scale is substantial, but its most important contribution may be conceptual rather than numerical. By manually curating and standardizing 59 datasets into a disease-progression-centered framework, the platform makes it possible to compare pathology across diseases and staging systems without starting from scratch each time.
That kind of infrastructure is especially valuable for researchers working at the intersection of single-cell biology, pathology, and computational analysis. It allows users to trace how cell states, tissue composition, and transcriptional programs evolve together. It also makes structured search and browsing possible across disease categories, which can accelerate hypothesis generation when investigators are looking for shared or divergent progression patterns.
PathoEvo is available at https://biomedlab-ai.com/pathoevo/. The abstract does not describe benchmark performance, user adoption metrics, or validation against external reference standards, so the current evidence base is limited to the resource description itself. Open questions remain about how consistently the stage mappings perform across all included diseases and whether future updates will expand the database beyond the 18 diseases and 27 progression frameworks already integrated.