aRrayLasso

aRrayLasso converts gene expression measurements between microarray platforms using Lasso-penalized generalized linear models to enable cross-platform integration and comparison.

Key Features:

• Implementation: Comprises five R functions for the core computational implementation.
• Data acquisition: Supports acquisition of gene expression data from public repositories such as Gene Expression Omnibus (GEO).
• Model training: Employs Lasso-penalized generalized linear models to train on probe-level data and handle high-dimensional microarray datasets.
• Platform mapping: Models relationships between individual probes across different microarray platforms rather than relying solely on probe set annotations or direct probe alignments.
• Prediction accuracy: Produces expression-level predictions with accuracy comparable to technical replicates of the same RNA pool.

Scientific Applications:

• Meta-Analyses: Combining datasets from multiple studies to increase statistical power and validate findings.
• Cross-Study Comparisons: Facilitating comparisons of gene expression profiles across studies that used different microarray technologies.
• Data Integration: Integrating diverse microarray datasets into unified analyses to expand biological interpretation.

Methodology:

Implements Lasso-penalized generalized linear models that impose an L1 penalty to promote sparsity and reduce overfitting, models relationships between individual probes across platforms, and is implemented as five R functions operating on gene expression data (e.g., from GEO).