STmiR: A Novel XGBoost-Based Framework for Spatially Resolved miRNA Activity Prediction in Cancer Transcriptomics

MicroRNAs (miRNAs) are critical regulators of gene expression in cancer biology; however, their spatial dynamics within tumor microenvironments (TME) remain underexplored owing to technical limitations in current spatial transcriptomics (ST) technologies. To address this gap, we present STmiR, a novel XGBoost-based framework for spatially resolved miRNA activity predictions. STmiR integrates bulk RNA-seq data (TCGA and CCLE) with spatial transcriptomics profiles to model nonlinear miRNA-mRNA interactions, achieving a high predictive accuracy (Spearman’s ρ > 0.8) across four major cancer types (breast, lung, ovarian, and prostate). Applied to 10X Visium ST datasets from nine cancers, STmiR identified six pan-cancer conserved miRNAs (e.g., hsa-miR-21, hsa-let-7a) consistently ranked in the top 40 across malignancies, and uncovered cell-type-specific regulatory networks in fibroblasts, B cells, and malignant cells. A breast cancer case study validated the utility of STmiR by linking miR-205 to androgen receptor (AR) signaling and miR-200b to epithelial-mesenchymal transition (EMT). By enabling the spatial mapping of miRNA activity, STmiR provides a transformative tool to dissect miRNA-mediated regulatory mechanisms in cancer progression and TME remodeling, with implications for biomarker discovery and precision oncology. Highlights First integration of XGBoost and spatial transcriptomics : STmiR pioneers a machine learning framework to predict miRNA activity in spatially heterogeneous tissues, overcoming limitations of linear correlation-based methods. High accuracy and generalizability : Demonstrates robust performance (Spearman’s ρ > 0.8) across four cancer types validated by independent datasets. Pan-cancer conserved miRNAs : Six miRNAs (e.g., hsa-miR-21 and hsa-let-7a) were shared across nine cancers, implicating their roles in core oncogenic pathways. Key Contributions Method: ological innovation : Combines XGBoost’s nonlinear modeling with spatial transcriptomics to resolve miRNA activity in multicellular contexts. Biological discovery : Uncovers the conserved and context-dependent roles of miRNAs in tumor progression and microenvironment crosstalk. Translational impact : Provides a computational platform for identifying spatially regulated therapeutic targets in cancer.