A clustering approach for localizing disease susceptibility loci

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Identifying trait loci is crucial for understanding complex genetic diseases.
• Simulated datasets offer controlled environments for testing genetic analysis methods.
• Previous methods may rely on specific genetic model assumptions.

Purpose of the Study:

• To evaluate the utility of Latent Class Analysis (LCA) and Cluster Analysis (CLA) for trait loci identification.
• To apply model-free statistical techniques to a simulated genetic disease dataset.
• To assess the performance of LCA and CLA in identifying disease susceptibility regions.

Main Methods:

• Latent Class Analysis (LCA) and Cluster Analysis (CLA) were employed.
• Non-overlapping subsets of affected relative pairs were created based on identity-by-descent (IBD) allele sharing.
• Subgroups with a high proportion of affected pairs were analyzed to locate trait loci.

Main Results:

• LCA and CLA successfully identified regions containing five out of seven simulated trait loci.
• The methods were applied without prior knowledge of the true disease model.
• Both affected and unaffected subjects' data were utilized in the model-free analyses.

Conclusions:

• LCA and CLA are effective model-free approaches for identifying trait loci in genetic studies.
• These methods demonstrate robustness in detecting disease susceptibility regions even without a known disease model.
• The findings support the use of LCA and CLA in genetic analysis, particularly with simulated data for method validation.