scPrognosis

scPrognosis predicts breast cancer prognosis by integrating single-cell RNA-sequencing (scRNA-seq)-derived EMT pseudotime and dynamic gene co-expression networks to identify prognostic gene signatures and apply them to bulk RNA-seq-based prediction.

Key Features:

• Single-Cell Resolution: Uses scRNA-seq data to capture intra-tumor heterogeneity at single-cell resolution.
• EMT Pseudotime Inference: Infers epithelial-to-mesenchymal transition (EMT) pseudotime for individual cells to characterize dynamic progression through EMT stages.
• Dynamic Gene Co-expression Network: Constructs dynamic gene co-expression networks that reflect changes in gene interactions across EMT stages.
• Integrative Model for Gene Selection: Selects EMT-related genes using an integrative model based on expression variation, differentiation across EMT pseudotime, and roles within the dynamic co-expression network.
• Prognosis Prediction Model: Applies the selected gene signatures as features in a prediction model built on bulk RNA-seq data for clinical prognosis.

Scientific Applications:

• Breast Cancer Prognosis: Enhances accuracy of prognostic prediction for breast cancer by incorporating single-cell EMT dynamics into bulk RNA-seq models.
• Mechanistic Insight: Provides insights into biological mechanisms linking EMT-associated dynamic gene expression changes to clinical outcomes.
• Generalizability: Can be applied to study other biological processes beyond EMT by leveraging scRNA-seq-derived dynamic signatures.

Methodology:

scPrognosis uses scRNA-seq data to infer EMT pseudotime for individual cells, constructs dynamic gene co-expression networks across EMT stages, employs an integrative model to select genes based on expression variation, differentiation across EMT pseudotime, and network roles, and uses the selected gene signatures as features in a prediction model built on bulk RNA-seq data.