Uncovering Functional Gene Regulatory Networks in Bulk and Single-Cell Data through Robust Transcription Factor...

Area of Science:

• Systems Biology
• Computational Biology
• Genomics

Background:

• Reconstructing genome-scale gene regulatory networks (GRNs) is challenging.
• Current methods estimating transcription factor activity (TFA) are sensitive to noise in prior knowledge.
• Accurate GRN inference requires robust TFA estimation.

Purpose of the Study:

• To develop a robust method for estimating TFA and inferring GRNs.
• To improve the accuracy and reliability of GRN reconstruction.
• To identify key regulators of the mouse embryonic stem cell state.

Main Methods:

• Developed MERLIN+P+TFA, a novel approach using prior knowledge-guided sparsity regularization.
• Applied the method to simulated and real expression data from yeast and mammalian systems.
• Validated inferred GRNs and prioritized key regulators in mouse embryonic stem cells.

Main Results:

• MERLIN+P+TFA robustly and accurately estimates TFA and improves GRN inference quality.
• The method shows improved GRN quality for both bulk and single-cell datasets.
• Experimentally validated 58 key regulators of the mouse embryonic stem cell state, identifying known and novel factors.

Conclusions:

• Regularized TFA estimation enhances various GRN inference algorithms.
• MERLIN+P+TFA offers a significant improvement for GRN reconstruction in systems biology.
• Inferred networks can precisely capture functional TF targets, highlighting the need for context-specific gold standards.