STmiR: A Novel XGBoost-based framework for spatially resolved miRNA activity prediction in cancer transcriptomics

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• MicroRNAs (miRNAs) are key gene regulators in cancer.
• Spatial dynamics of miRNAs within tumor microenvironments (TMEs) are poorly understood due to technological limitations.

Purpose of the Study:

• To introduce STmiR, a novel framework for predicting spatially resolved miRNA activity.
• To model nonlinear miRNA-mRNA interactions within the TME.

Main Methods:

• Developed STmiR, an XGBoost-based framework integrating bulk RNA-seq (TCGA, CCLE) and spatial transcriptomics (ST) data.
• Modeled miRNA-mRNA interactions to predict miRNA activity with high accuracy (Spearman's ρ > 0.8).
• Validated performance against experimental miRNA expression data.

Main Results:

• STmiR successfully predicted miRNA activity across four major cancer types.
• Identified six conserved pan-cancer miRNAs and uncovered cell-type-specific regulatory networks.
• Demonstrated utility in a breast cancer case study, revealing miRNA-target relationships linked to cancer pathways.

Conclusions:

• STmiR enables spatial mapping of miRNA activity, offering a transformative tool for cancer research.
• Facilitates dissection of miRNA-mediated mechanisms in cancer progression and TME remodeling.
• Has implications for biomarker discovery and precision oncology.