scMINER

scMINER enables mutual information-based clustering, network inference, and hidden driver identification from single-cell transcriptomic data.

Key Features:

• Unsupervised Cell Clustering: scMINER clusters cells using mutual information-based methods, demonstrating superior accuracy in distinguishing closely related cell populations compared to five state-of-the-art algorithms.
• Transcription Factor Network Inference: scMINER infers transcription factor networks and identifies key transcription factor drivers, outperforming three established methods including SCENIC as validated by ATAC-seq and CROP-seq.
• Signaling Protein Network Inference: scMINER infers signaling protein networks and drivers with high accuracy, providing an advantage in multimodal single cell data analysis.
• Hidden Driver Identification: scMINER identifies hidden drivers from single-cell transcriptomic data, including drivers involved in T cell exhaustion and Treg tissue specification.

Scientific Applications:

• T Cell Exhaustion Research: scMINER reveals key transcription factor drivers involved in T cell exhaustion.
• Treg Tissue Specification Research: scMINER identifies transcription factor drivers involved in Treg tissue specification.
• Multimodal Single Cell Data Analysis: scMINER enables inference of signaling protein networks and drivers for multimodal single cell data analysis.
• Closely Related Cell Population Distinguishing: scMINER excels in distinguishing closely related cell populations from single-cell RNA-seq data.

Methodology:

scMINER uses mutual information-based methods for unsupervised cell clustering, network inference, and driver identification from single-cell transcriptomic data. Validation was performed using ATAC-seq and CROP-seq.