DGH-GO

DGH-GO stratifies putative disease-causing genes by Gene Ontology-based functional similarity to identify clusters that dissect genetic heterogeneity in complex diseases.

Key Features:

• Stratification of Disease-Causing Genes: Stratifies putative disease-causing genes into clusters to clarify how distinct gene groups contribute to different disease outcomes.
• Semantic Similarity Matrix Creation: Utilizes Gene Ontology (GO) to construct a semantic similarity matrix for input genes.
• Dimension Reduction Visualization: Projects the similarity matrix into 2D using t-SNE, Principal Component Analysis (PCA), UMAP, and Principal Coordinate Analysis (PCoA).
• Advanced Clustering Methods: Applies K-means, Hierarchical, Fuzzy, and Partitioning Around Medoids (PAM) clustering to identify functionally similar gene groups.
• Clustering Parameter Adjustment: Provides tunable clustering parameters to explore alternative gene stratifications.

Scientific Applications:

• Autism Spectrum Disorder (ASD) analysis: Applied to ASD where it identified four distinct gene clusters associated with different biological mechanisms and clinical outcomes.
• Cross-NDD gene aggregation and shared etiologies: Used to analyze genes shared among neurodevelopmental disorders (NDDs), revealing aggregation of shared genes into similar clusters that suggest potential shared etiologies.

Methodology:

Computes Gene Ontology-based semantic similarity matrices, applies dimensionality reduction (t-SNE, PCA, UMAP, PCoA), and performs clustering with K-means, Hierarchical, Fuzzy, and Partitioning Around Medoids (PAM).