contamDE-lm

contamDE-lm applies a novel linear model to next-generation RNA-seq data to perform differential gene expression analysis while adjusting for cellular contamination in paired tumor–normal samples.

Key Features:

• Accounting for Cellular Contamination: Accounts for cellular contamination in tumor RNA-seq by incorporating gene-wise information to reduce individual residual variances and mitigate contamination-driven false positives and false negatives.
• Novel Linear Model: Implements a linear-model framework that addresses tumor–normal correlations in paired samples to mitigate bias and improve reliability of DE results.
• Computational Efficiency: Provides computational speed advantages for large RNA-seq datasets with comparisons noted against limma and DESeq2.
• Statistical Robustness: Manages contamination-induced noise to improve detection of differentially expressed genes between normal and tumor samples and among tumor subtypes.
• Sequencing and Design Suitability: Designed for next-generation RNA-seq data and paired tumor–normal study designs.

Scientific Applications:

• Cancer differential expression analysis: Analyzes differential gene expression between tumor and adjacent normal tissues to provide insights into tumorigenesis and potential therapeutic targets.
• Heterogeneous tissue studies: Applies to RNA-seq datasets from heterogeneous or contaminated tissue samples to improve DE inference in presence of mixed cell populations.

Methodology:

Applies a novel linear-model approach that integrates gene-wise information to adjust for cellular contamination, reduce residual variances, and account for tumor–normal correlations in paired RNA-seq differential expression analyses; it has been applied to simulated data and to TCGA and GEO datasets.