Multivariate sib-pair linkage analysis of longitudinal phenotypes by three step-wise analysis approaches

Area of Science:

• Genetics
• Statistical Genetics
• Computational Biology

Background:

• Complex diseases require advanced statistical methods for genetic linkage analysis.
• Existing methods include linear models, generalized linear models, and data mining techniques.
• Novel pattern recognition techniques offer potential improvements.

Purpose of the Study:

• To investigate the utility and properties of step-wise discriminant analysis for complex disease genetic mapping.
• To compare step-wise discriminant analysis with step-wise logistic and linear regression.
• To assess the potential for detecting gene-gene and gene-environment interactions.

Main Methods:

• Utilized chromosome 10 linkage data from the Framingham Heart Study.
• Employed step-wise discriminant analysis, step-wise logistic regression, and step-wise linear regression.
• Compared methods based on statistical significance and gene localization accuracy.

Main Results:

• Step-wise discriminant analysis demonstrated superior performance in statistical significance and gene localization.
• All three methods successfully identified the hypertension-linked chromosomal region (marker GATA64A09).
• Evidence of gene x environment interactions (e.g., GATA64A09 and hypertension treatment) and gene x gene interactions (e.g., GATA64A09 and GATA115E01) was found.

Conclusions:

• Step-wise discriminant analysis is confirmed as a valuable method for the genetic mapping of complex diseases.
• The pattern recognition technique shows promise for investigating gene x gene and gene x environment interactions.
• Discriminant sib-pair linkage analysis may offer greater power for binary phenotypes compared to trichotomous ones.