Phenomizer

Phenomizer identifies and ranks candidate hereditary diseases from clinical phenotypes by integrating phenotype-gene associations with human disease networks and protein-protein interaction (PPI) networks.

Key Features:

• Integration of Heterogeneous Networks: Integrates human disease networks, protein-protein interaction (PPI) networks, and gene-disease bipartite networks to analyze relationships among phenotypes, genes, and diseases.
• Phenotype-Gene Association: Associates clinical phenotypes with causative genes to construct a phenotype-gene network.
• Path Frequency Analysis: Counts frequencies of all possible paths from a phenotype to a disease through associated genes to construct a phenotype-disease bipartite (PDB) network.
• Candidate Disease Ranking: Generates candidate edge weights in the PDB network and ranks potential diseases by association strength, with ranking performance evaluated using Normalized Discounted Cumulative Gain (NDCG).

Scientific Applications:

• Differential diagnosis in human genetics: Provides a data-driven approach to suggest plausible hereditary diseases that explain observed clinical phenotypes.
• Network-based disease and gene prioritization: Enables prioritization of candidate diseases and implicated genes by integrating phenotype-gene associations and network path analyses.

Methodology:

Constructs a phenotype-disease bipartite (PDB) network by associating phenotypes with genes and counting path frequencies from phenotypes to diseases through those genes to generate weighted phenotype-disease edges, then ranks diseases and evaluates ranking quality using NDCG.