Cosbin

Cosbin normalizes gene expression data across biologically diverse samples to correct normalization bias arising from asymmetric differential expression and to enable accurate inference and comparability.

Key Features:

• Cosine score-based iterative normalization: Uses cosine scores to evaluate cross-condition expression patterns and iteratively identify and remove asymmetrically differentially expressed genes that bias normalization.
• Identification of consistently expressed genes: Iteratively refines the dataset to identify genes with consistent expression across conditions to serve as normalization references.
• Sample-wise normalization factors: Computes sample-wise normalization factors from consistently expressed genes to adjust gene expression measures across samples and conditions.
• Performance validation: Demonstrates enhanced performance versus six representative peer methods in simulation studies and analyses of real multi-omics expression datasets, addressing normalization bias caused by asymmetric differential expression.

Scientific Applications:

• Multi-omics data normalization: Normalizes multi-omics expression datasets across diverse phenotypic groups to enable detection of true molecular signals.
• Genomics and transcriptomics differential expression: Improves comparability for detecting differential expression in genomics and transcriptomics studies with significant asymmetry.
• Personalized medicine: Enhances identification of biologically relevant molecular signals that inform personalized medicine analyses.

Methodology:

Calculates cosine scores to assess cross-condition expression patterns; iteratively eliminates asymmetrically differentially expressed genes; identifies consistently expressed genes and computes sample-wise normalization factors from those genes.