ENDEAVOUR

ENDEAVOUR prioritizes candidate genes by integrating models inferred from heterogeneous genomic data to identify genes associated with biological processes and diseases.

Key Features:

• Training set-based prioritization: Users provide a set of training genes known to be involved in the target biological process or disease and candidate genes for prioritization.
• Model inference from heterogeneous genomic datasets: Models are inferred from ontologies, annotations, protein-protein interactions, cis-regulatory information, gene expression datasets, sequence information, and text-mining data.
• Model application to candidates: Each inferred model is applied to candidate genes to rank them against the profile of the training genes.
• Integration of rankings: Individual model rankings are merged into a global ranking using order statistics.
• Multi-species support: Supports Homo sapiens, Mus musculus, Rattus norvegicus, and Caenorhabditis elegans.
• Flexibility in data sources: Additional data sources can be incorporated into the inference and ranking process.

Scientific Applications:

• Complex disease and pathway gene prioritization: Applied to prioritize candidate genes in studies of complex diseases and biological pathways.
• Obesity and Type II diabetes: Used to prioritize candidate genes associated with obesity and Type II diabetes.
• Cleft lip and palate: Used to prioritize candidate genes associated with cleft lip and palate.
• Pulmonary fibrosis: Used to prioritize candidate genes associated with pulmonary fibrosis.
• Myeloid differentiation: Applied to identify genes involved in myeloid differentiation.
• Craniofacial development and DiGeorge-like defects: Identified a novel gene within a 2-Mb chromosomal region linked to DiGeorge-like birth defects affecting craniofacial development.
• Benchmark and experimental validation: In silico benchmarking and experimental validation prioritized 627 genes in disease datasets and 76 genes in biological pathway sets.

Methodology:

Models are inferred from multiple heterogeneous genomic data sources, applied to candidate genes to produce model-specific rankings, and merged into a global ranking using order statistics; additional data sources can be incorporated.