Dissecting Glioblastoma Risk Signatures in the Tumor Immune Microenvironment Based on Multi-dimensional Transcriptomics

Research summary

Glioblastoma (GBM) is characterized by pronounced tumor heterogeneity and a complex immune microenvironment, contributing to poor patient survival outcomes. In this study, we comprehensively dissected the tumor microenvironment (TME) and uncovered potential molecular mechanisms by integrating single-cell, bulk, and spatial transcriptomic data.

Hallmarks of malignancy and cell cycle regulatory pathways were consistently enriched across these modalities. We identified seven hallmark-related prognostic signatures (HMsig) using machine learning algorithms—namely AEBP1, ASF1A, PRPS1, DCC, OPHN1, IL13RA2, and HDAC5—whose importance in predicting patient outcomes was validated through SHAP algorithm analysis.

Ligand-receptor (LR) interaction analysis revealed that interactions involving OPHN1 were associated with poorer prognosis. Additionally, Immune checkpoint genes (ICG) LAG3, PDCD1, and HAVCR2 were found to be substantially upregulated along the pseudotime trajectory of T-cell progression. Spatial transcriptomic analysis consistently demonstrated the existence of synergistic gene interactions, deciphering the immunomodulatory functions of GBM biomarkers in the TME.

Repository Structure

The code is organized into sequential steps corresponding to the analysis workflow in the manuscript.

GBM-TME-Analysis/
├── README.md                 # Project Overview & Instructions
├── LICENSE                   # MIT License
└── Code_summary/                  # Analysis Source Code
    ├── Step1_Single-cell RNA-seq Analysis Pipeline.R      # QC, Clustering, Annotation, InferCNV, Hallmarks of scRNA-seq
    ├── Step2_Bulk RNA-seq Analysis Pipeline.R       # Batch Correction, Consensus Clustering, Hallmarks of bulk RNA-seq
    ├── Step3_Model Construction and Validation.R    # ML Model (StepCox+RSF), SHAP, HMsig (7 genes)
    ├── Step4_Cell-Cell Commnication Analysis.R            # L-R Interactions
    ├── Step5_Trajectory Analysis.R          # Monocle3 (T-cell trajectory)
    ├── Step6_Regulon Analysis.R              # Regulon Activity & Clinical Outcome
    └── Step7_Spatial Transcriptomics Analysis.R             # Spatial Deconvolution (CARD) for Validation