weitrix

weitrix provides data types and tools for working with matrices having precision weights and missing data, enabling analysis of many types of high-throughput data with a common representation compatible with base R function lm and the package limma.

Key Features:

• Weight Calibration: Calibrates weights to account for known predictors of precision.
• Excess Variability Detection: Identifies rows with excess variability in the data.
• Differential Testing: Performs differential testing and finds rows with the largest confident differences.
• PCA-like Component Extraction: Finds PCA-like components of variation even with many missing values, rotated so that individual components may be meaningfully interpreted.
• Scalable Data Support: Supports DelayedArray matrices and BiocParallel for efficient computation.

Scientific Applications:

• High-Throughput Data Analysis: Provides a common representation applicable to many types of high-throughput data with precision weights and missing values.
• Weighted Linear Modeling: Enables analyses compatible with base R function lm and the limma package for precision-weighted data.
• Variability Assessment: Identifies features showing excess variability beyond expected precision.
• Pattern Discovery: Extracts interpretable components of variation from sparse or incomplete datasets.

Methodology:

Weight calibration accounts for known predictors of precision; excess variability detection, differential testing, and PCA-like component extraction with rotation for interpretability are performed on precision-weighted matrices with missing data.