Gene circuits: building gene-regulatory networks from single-cell data

Inferring cell-type-specific gene-regulatory networks from single-cell data and perturbing them in silico.

Network inference with scRNA-seq: from tissue sample to count matrix, clustering into cell types, and a gene-regulatory network per cell type; disease and healthy networks are then compared.
Adapted from Figures 1b and 3 of Cha J, Lee I. Single-cell network biology for resolving cellular heterogeneity in human diseases. Exp Mol Med (2020). Licence: CC BY 4.0.

Background

Biological systems are complex: genes act through their interactions with one another, and these interactions can be represented as gene-regulatory networks (GRNs). Single-cell data make it possible to build such a network for each cell type.

Two questions follow: how such networks can be created from sparse data, and how perturbation biology approaches and in silico gene knock-outs (node elimination) can then be used to understand human health. Methods such as SCENIC and CellOracle address these steps.

Questions we are working on

  • How do existing approaches, for example SCENIC and CellOracle, construct GRNs from single-cell data, and how should the resulting networks be compared?
  • How can GRNs be perturbed, for example with CellOracle?
  • What are the bottlenecks of current GRN inference and perturbation methods?
  • Which similar methods can be built on flow matching?

Key literature

  1. Cha J, Lee I. Single-cell network biology for resolving cellular heterogeneity in human diseases. Exp Mol Med (2020).
  2. Kamimoto K, et al. Dissecting cell identity via network inference and in silico gene perturbation. Nature (2023). [CellOracle]
  3. Aibar S, et al. SCENIC: single-cell regulatory network inference and clustering. Nat Methods (2017).