NITUMID

NITUMID deconvolves bulk RNA gene expression data using Nonnegative Matrix Factorization to simultaneously estimate tumor and immune cell proportions and detect clinical and prognostic signals in tumor microenvironments.

Key Features:

• Nonnegative Matrix Factorization Framework: Utilizes Nonnegative Matrix Factorization (NMF) to deconvolute complex RNA data into tumor and immune components within the tumor microenvironment.
• Simultaneous Estimation of Cell Proportions: Estimates the proportions of tumor and immune cells concurrently from bulk RNA data.
• Accommodation for Variable mRNA Levels: Accounts for differences in mRNA expression levels across cell types to improve accuracy of proportion estimates.
• Enhanced Profiling Accuracy: Demonstrates superior cell type profiling accuracy compared with existing deconvolution methods via simulations and real data analyses.
• Clinical and Prognostic Signal Detection: Identifies clinical and prognostic signals from gene expression profiles to inform therapeutic response prediction and biomarker discovery.

Scientific Applications:

• Tumor microenvironment profiling: Characterizes the composition of tumor and immune cells in TMEs for oncological research.
• Therapeutic response prediction: Provides cellular composition-based signals that can be used to assess or predict responses to therapies.
• Biomarker and prognostic signal identification: Detects prognostic signals and potential biomarkers for cancer progression and treatment response from gene expression profiles.

Methodology:

Implements a Nonnegative Matrix Factorization algorithm in R and validates performance using comprehensive simulations and analyses of real data sets.