Mining human phenome to investigate modularity of complex disorders

Area of Science:

• Genetics and Bioinformatics
• Computational Biology
• Medical Informatics

Background:

• Biomedical research aims to understand factors controlling clinical disease phenotypes.
• Identical mutations can cause varied phenotypes due to genetic and environmental interactions.
• Different diseases often share extensive phenotypic features.

Purpose of the Study:

• To test if disease similarities/differences reflect "disease phenotype modules" activation.
• To develop a systematic method for analyzing disease-phenotype relationships.
• To identify non-trivial associations among disease entities.

Main Methods:

• Systematic parsing of Online Mendelian Inheritance in Man (OMIM) and Syndrome DB databases.
• Utilizing the Unified Medical Language System (UMLS) for feature extraction.
• Constructing a disease-clinical phenotypic feature matrix for clustering algorithms.

Main Results:

• Cardiovascular Syndromes were used as a model system.
• Demonstrated the importance of phenotypic generalization and specificity.
• Enabled retrieval of non-trivial associations among disease entities.

Conclusions:

• Phenotypic similarities and differences can be represented by "disease phenotype modules".
• This approach facilitates the discovery of shared protein domains and pathway functions of causal genes.
• Provides a framework for understanding complex disease relationships and genetic underpinnings.