HARP integrates ~9 million PBMCs with 192 immune cell annotations and uncovers a sex dimorphism in prostaglandin signaling

The work presents HARP, an integrated atlas of ~9 million PBMCs from >2,600 donors across 15 studies spanning four continents, neonates to 97 years, and health and diverse immune-related diseases. Previously, the absence of a PBMC reference with standardized immune cell annotations constrained comparisons between scRNAseq studies; HARP addresses this gap through unified integration. Based on the scale stated in the abstract: ~9 million cells, >2,600 donors, 15 studies, four continents, ages from neonates to 97 years.

The study developed optimized integration workflows, including novel label-free metrics to assess integration quality, and generated community-driven consensus annotations for 192 immune cell subsets, identifying rare populations representing as few as 0.004% of PBMCs. It advances annotation granularity to 192 subsets including extremely rare cells, while providing a label-free measure of integration quality. The abstract explicitly states 192 subsets, the 0.004% rare-population figure, and the label-free metric as a methodological element.

HARP reveals coordinated cellular modules associated with age, sex, and disease, identifies female-biased interferon and inflammatory gene programs and their perturbation by COVID-19, and uncovers a sexual dimorphism in prostaglandin signaling. It extends the reference from an annotation tool to a resource for discovering cross-population biological signals, particularly sex-related immune differences. The abstract reports these findings as statements without specific effect sizes or statistical details.

The study introduces scTiger, a hierarchical label-transfer framework that accurately projects the 192 cell annotations onto ~18 million additional PBMCs. It makes the reference annotations transferable to new data, offering a reusable tool for harmonizing future PBMC studies. The abstract gives the scale of ~18 million additional PBMCs projected and describes the projection as accurate, without listing specific accuracy values.

The work targets researchers who need harmonized PBMC annotation and cross-study comparison, and applies to scRNAseq data from healthy and immune-related disease populations; its stated role is as a basis for harmonizing future PBMC studies and for emerging clinical applications of single-cell genomics.

The reading scope here is incomplete, containing only the abstract and competing-interest statement, without figures, methodological details, or statistical results; therefore the specific performance of the integration-quality metrics, the basis for the 192 annotations, the effect sizes of the sex differences and COVID-19 perturbation, and the quantitative evidence for scTiger projection accuracy cannot be assessed. These are open questions a reader would need to confirm in the full text.