ShareNet

ShareNet enhances inference of cell type-specific gene regulatory networks from single-cell RNA-sequencing (scRNA-seq) data by using a Bayesian information-sharing framework to identify unique regulatory associations across cell types and rare transcriptional states.

Key Features:

• Bayesian framework: Implements a Bayesian model to improve estimates of gene regulatory networks.
• Information-sharing structure: Shares information across related cell types to propagate relevant signals between networks.
• Adaptive optimization: Uses adaptive optimization to refine initial network estimates and boost precision of regulatory associations.
• Algorithm-agnostic input: Accepts initial network estimates generated by any general network inference algorithm.
• Rare-state support: Improves inference for rare transcriptional states with limited sample sizes.
• Cross-context sensitivity: Captures regulatory network rewiring across cell types, tissues, or dynamic processes.
• Revised-network output: Produces revised networks that better reflect cell type-specific gene regulatory relationships.

Scientific Applications:

• Cell type-specific GRN inference: Inferring gene regulatory networks specific to individual cell types from scRNA-seq data.
• Detection of unique regulatory associations: Identifying regulatory links that are specific to particular cell types or conditions.
• Analysis of rare transcriptional states: Enhancing network inference where conventional methods fail due to small sample sizes.
• Comparative regulatory analysis: Revealing regulatory rewiring across cell types, tissues, or dynamic biological processes, including benchmarking on scRNA-seq datasets.

Methodology:

ShareNet takes initial network estimates from any chosen inference algorithm as input and applies a Bayesian information-sharing model with adaptive optimization across related cell types to produce revised networks.