GEMDeCan

GEMDeCan deconvolves bulk tumor RNAseq and DNA methylation data to estimate immune and cancer cell-type proportions and profile the tumor immune microenvironment for immunotherapy response analyses.

Key Features:

• Complementary Deconvolution Approaches: Integrates gene expression and DNA methylation data to construct signature matrices that estimate proportions of immune cells and cancer cells from RNAseq and DNA methylation profiles.
• Advanced Signature Matrices: Provides five deconvolution signature matrices tested on in-silico and in-vitro mixtures, peripheral blood samples, TCGA cancer samples, and single-cell melanoma datasets, showing comparable or superior correlation with FACS measurements and cancer purity estimates.
• Incorporation of 3D Chromatin Structure Data: Uses publicly available 3D chromatin structure data from hematopoietic cells to expand RNAseq signature matrices by including genes associated with methylated CpGs in promoters or genomic regions contacting those promoters, improving matrix performance.
• Predictive Capability for Immunotherapy Response: Demonstrates prediction of patient responses to immune checkpoint inhibitors across three melanoma cohorts using bulk tumor gene expression data.
• Snakemake Analysis Pipeline: Implements a Snakemake pipeline to process analyses from raw sequencing data to deconvolution and to apply various gene expression signature matrices for bulk RNASeq and DNA methylation data.

Scientific Applications:

• Tumor Immune Microenvironment Profiling: Estimates cell-type composition in tumor biopsies to characterize immune and cancer cell distributions using RNAseq and DNA methylation signatures.
• Benchmarking and Validation: Validates deconvolution outputs against FACS measurements and cancer purity estimates across datasets including TCGA and single-cell melanoma data.
• Immunotherapy Response Prediction: Supports prediction of responses to immune checkpoint inhibitors in melanoma cohorts using bulk tumor gene expression data.

Methodology:

Constructs deconvolution signature matrices by integrating gene expression and DNA methylation data; validates matrices against FACS measurements and cancer purity estimates; expands RNAseq signature matrices using 3D chromatin contacts to include genes linked to methylated CpGs in promoters or contacting regions in hematopoietic cells; provides a Snakemake pipeline to run analyses from raw sequencing data to deconvolution.