scRAD supports a reproducibility-driven framework for assessing the expression differences in single-cell RNA-Seq (scRNA-Seq) data collected from multiple donors. scRAD provides tools for:

  • Reproducible gene module analysis – identifying axes of variation common across donors
  • Irreproducible Discovery Rate (IDR) analysis (Q. Li et al. 2011) for assessing signal reproducibility across multiple donors
  • Other meta-analysis applicable to multi-donor studies

As single-cell technologies mature and push the boundaries of study resolution and scale, they promise to enrich our understanding of several complex biological systems. In the context of human disease, studies of single donors may prove very fruitful in the characterization and cataloging of diverse cell states. While single-subject studies can be used to generate bold hypotheses, these approaches do not guarantee results applicable to other individuals with one or more related phenotypes. More generally, if the goal of an analysis is to predict phenomena linked to a specific phenotype – in humans, mice, or other models with unknown or uncontrolled sources of variation – it is important to prioritize reproducibility across multiple sources. We developed scRAD to identify conserved trends in single-cell mRNA expression across samples. Please visit the following links to access scRAD code and accompanying data:

Github Repository:

https://github.com/YosefLab/scRAD

GEO Data:

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE108445

References

Q. Li, J. B. Brown, H. Huang, P. J. Bickel, Measuring reproducibility of high-throughput experiments. Ann. Appl. Stat. 5, 1752-1779 (2011).

Lab Members Involved

Alex K. Shalek Alex K. Shalek
José Ordovas-Montañes José Ordovas-Montañes
Kellie Kolb Kellie Kolb
Sam Kazer Sam Kazer