The visualization of Orphadata neurology phenotypes.

Area of Science:

• Genetics and Bioinformatics
• Computational Biology
• Medical Informatics

Background:

• Disease phenotypes, comprising observable signs and patient-reported symptoms, are crucial for diagnosis.
• Large repositories like Online Mendelian Inheritance in Man (OMIM) and Orphadata contain extensive phenotype data, particularly for neurologic diseases.
• Current phenotype data lacks effective visualization, hindering comparative analysis.

Purpose of the Study:

• To develop and demonstrate a novel visualization technique for neurologic disease phenotypes.
• To address the limitation of non-visualized concept lists in existing phenotype repositories.
• To enable side-by-side comparisons of complex disease data.

Main Methods:

• Utilized subsumption to reduce 2,946 phenotype features into 30 superclasses.
• Converted variable-length phenotype feature lists into fixed-length vectors.
• Aggregated phenotype vectors into matrices for visualization as heat maps and word clouds.

Main Results:

• Successfully visualized neuro-phenotypes of 32 dystonic diseases from Orphadata.
• Demonstrated that subsumption effectively collapses phenotype features into manageable superclasses.
• Generated heat maps for matrix-based disease comparisons and word clouds for individual disease visualization.

Conclusions:

• The developed method provides effective visualization of complex neurologic disease phenotypes.
• Heat maps and word clouds facilitate intuitive disease comparison and understanding.
• This approach enhances the utility of large phenotype repositories for research.