Cell
Volume 174, Issue 3, 26 July 2018, Pages 716-729.e27
Journal home page for Cell

Resource
Recovering Gene Interactions from Single-Cell Data Using Data Diffusion

https://doi.org/10.1016/j.cell.2018.05.061Get rights and content
Under an Elsevier user license
open archive

Highlights

  • MAGIC restores noisy and sparse single-cell data using diffusion geometry

  • Corrected data are amenable to myriad downstream analyses

  • MAGIC enables archetypal analysis and inference of gene interactions

  • Transcription factor targets can be predicted without perturbation after MAGIC

Summary

Single-cell RNA sequencing technologies suffer from many sources of technical noise, including under-sampling of mRNA molecules, often termed “dropout,” which can severely obscure important gene-gene relationships. To address this, we developed MAGIC (Markov affinity-based graph imputation of cells), a method that shares information across similar cells, via data diffusion, to denoise the cell count matrix and fill in missing transcripts. We validate MAGIC on several biological systems and find it effective at recovering gene-gene relationships and additional structures. Applied to the epithilial to mesenchymal transition, MAGIC reveals a phenotypic continuum, with the majority of cells residing in intermediate states that display stem-like signatures, and infers known and previously uncharacterized regulatory interactions, demonstrating that our approach can successfully uncover regulatory relations without perturbations.

Keywords

manifold learning
imputation
single-cell RNA sequencing
EMT
regulatory networks

Cited by (0)

9

Senior author

10

Lead Contact