eisa

eisa implements the Iterative Signature Algorithm (ISA) and the Progressive Iterative Signature Algorithm (PISA) to perform biclustering and identify overlapping transcription modules in gene expression and other tabular biological data.

Key Features:

• Biclustering: Identifies correlated blocks or transcription modules within gene expression and other tabular biological datasets.
• Overlap detection: Detects overlapping modules allowing genes and samples to belong to multiple modules.
• Noise robustness: Demonstrates resilience against noise in expression data to extract meaningful patterns.
• Progressive elimination (PISA): Incorporates PISA's sequential module-elimination process to enable unsupervised detection of both large and small regulatory modules.
• Unsupervised identification: Performs module discovery without requiring predefined labels or supervision.
• Gene Ontology reference: Supports evaluation of identified modules using the Gene Ontology database as a reference framework.
• Bioconductor compatibility: Operates with standard BioConductor data structures.

Scientific Applications:

• Regulatory module discovery: Identification of regulatory modules in gene-expression datasets, including microarray data.
• Gene regulatory network analysis: Analysis of complex gene regulatory networks where overlapping modules are common.
• Yeast expression studies: Applied to large yeast gene-expression datasets to extract biologically relevant modules.
• Comparative evaluation: When referenced to Gene Ontology, can outperform methods based on high-throughput transcription-factor binding experiments or comparative genomics for module identification.

Methodology:

Performs biclustering via the Iterative Signature Algorithm (ISA); applies the Progressive ISA (PISA) sequential module-elimination procedure for iterative removal of identified modules; conducts unsupervised identification of regulatory modules and can use the Gene Ontology database for reference-based evaluation.