Characteristic Direction Method

Characteristic Direction identifies differentially expressed genes using a geometrical multivariate approach to increase sensitivity in microarray and RNA-Seq analyses of transcription factor and drug perturbation experiments.

Key Features:

• Geometrical multivariate approach: Leverages geometric principles in a multivariate framework to detect differential expression.
• Improved sensitivity: Demonstrates higher sensitivity compared to univariate methods such as Significance Analysis of Microarrays (SAM), Linear Models for Microarray Data (LIMMA), DESeq, and edgeR.
• Transcription factor and drug perturbation focus: Applied to transcription factor perturbation and drug response microarray experiments.
• Benchmarking on synthetic and RNA-Seq data: Benchmarked using synthetic datasets and real-world RNA-Seq data.
• GEO dataset evaluation: Evaluated on 73 transcription factor perturbation experiments and 130 drug perturbation experiments from the Gene Expression Omnibus (GEO).
• Validation in STAT3 DLBCL study: Validated using an RNA-Seq study profiling genome-wide gene expression and STAT3 DNA binding in diffuse large B-cell lymphoma subtypes.
• Drug-target interactome association: Identifies differentially expressed genes associated with proteins interacting with drug targets across cellular contexts.
• Gene set enrichment enhancement: Enhances gene set enrichment analyses and outperforms Gene Set Enrichment Analysis (GSEA) and hypergeometric tests.

Scientific Applications:

• Transcription factor perturbation analysis: Identifying DEGs from transcription factor perturbation experiments in microarray and RNA-Seq datasets.
• Drug response profiling: Detecting gene expression changes associated with drug perturbations and linking DEGs to drug-target interactors.
• Pathway and mechanism discovery: Revealing biological processes and pathways that may be undetected by traditional DEG methods.
• Gene set enrichment improvement: Improving sensitivity and relevance of gene set enrichment analyses compared to GSEA and hypergeometric tests.
• Benchmarking and validation: Supporting method benchmarking with synthetic datasets and validation using specific RNA-Seq studies such as the STAT3 DLBCL profiling.

Methodology:

Leverages geometrical principles in a multivariate framework to discern patterns in gene expression data that indicate differential expression.