plgem

plgem models power-law variance-versus-mean dependence in microarray and proteomics data and uses model-derived measurement spread estimates with a resampling-based hypothesis testing algorithm to identify differentially expressed genes, including for Affymetrix GeneChip datasets.

Key Features:

• Empirical variance-mean modeling: Models the dispersion of repeated measures as a power-law relationship between variance and mean to capture measurement variability.
• Affymetrix GeneChip applicability: Tailored and validated for Affymetrix GeneChip datasets using both proprietary and publicly available data.
• Model-derived spread estimates: Derives measurement spread estimates from the power-law model and incorporates them into downstream testing.
• Resampling-based hypothesis testing: Uses a resampling-based algorithm for hypothesis testing to detect differentially expressed genes.
• Robustness to replicate number: Demonstrates consistent performance across varying numbers of replicates, including single-sample conditions.

Scientific Applications:

• Differential expression analysis: Identification of differentially expressed genes in microarray and proteomics experiments.
• Transcriptomic studies of disease and intervention: Detection of transcriptional modulations associated with physiological states, diseases, or interventions.
• Genomics and molecular biology research: Support for biomarker discovery and studies in genomics, molecular biology, and personalized medicine.

Methodology:

Empirical modeling of variance-versus-mean dependence using a power law; derivation of model-based measurement spread estimates; resampling-based hypothesis testing for DEG detection; validation on Affymetrix GeneChip datasets (proprietary and public).