STGRNS: an interpretable transformer-based method for inferring gene regulatory networks from single-cell...

Area of Science:

• Systems Biology
• Computational Biology
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables the inference of cell-specific gene regulatory networks (GRNs).
• Inferring GRNs from scRNA-seq data presents challenges due to cellular heterogeneity.
• Existing methods struggle to fully address the complexities of scRNA-seq data.

Purpose of the Study:

• To develop an interpretable method for inferring GRNs from scRNA-seq data.
• To overcome limitations in current GRN inference techniques, particularly concerning cellular heterogeneity.
• To provide a more accurate and explainable approach for understanding gene regulation at the single-cell level.

Main Methods:

• Developed STGRNS, an interpretable transformer-based algorithm for GRN inference.
• Introduced a gene expression motif technique to convert gene pairs into sub-vectors for transformer input.
• Utilized a transformer encoder architecture to process gene expression data.

Main Results:

• STGRNS demonstrated superior performance compared to popular GRN inference methods.
• The method achieved high accuracy across extensive benchmark datasets, including static and time-series scRNA-seq data.
• STGRNS proved to be more interpretable than "black box" deep learning methods.

Conclusions:

• STGRNS is an effective and interpretable tool for inferring GRNs from scRNA-seq data.
• The gene expression motif technique enhances the accuracy of GRN inference, especially for heterogeneous datasets.
• The developed method offers a significant advancement in systems biology for understanding gene regulation.