GLDADec

GLDADec performs bulk gene expression deconvolution using a guided latent Dirichlet allocation (LDA) modeling approach that leverages cell type-specific marker genes to estimate topic distributions for each sample.

Key Features:

• Marker-Guided Topic Modeling: GLDADec guides topics using cell type-specific marker gene names as partial prior information to estimate topic distributions for each sample.
• High Estimation Performance: Benchmarking using blood-derived datasets demonstrates high estimation performance and robustness compared to existing methods.
• Biological Interpretability Evaluation: The method evaluates biological interpretability by examining enrichment of biological processes for topics.
• Data-Driven Cell Type Analysis: GLDADec enables comprehensive cell type analysis in a data-driven manner for heterogeneous tissue bulk data.
• Clinical Utility: The approach enables subtype stratification and survival analysis based on estimated cell type proportions when applied to The Cancer Genome Atlas tumor samples.

Scientific Applications:

• Immunology Research: GLDADec infers cell type proportions from bulk transcriptome data for immunological studies.
• Oncology Research: The method enables analysis of tumor microenvironment composition and cell type proportions in cancer research.
• Heterogeneous Tissue Analysis: GLDADec performs comprehensive cell type analysis on heterogeneous tissue bulk data in a data-driven manner.
• Clinical Stratification and Prognosis: The tool supports subtype stratification and survival analysis based on estimated cell type proportions in clinical tumor samples.

Methodology:

GLDADec uses guided latent Dirichlet allocation (LDA) modeling with cell type-specific marker gene names as partial prior information to guide topics and estimate topic distributions for each sample.