A tobit variance-component method for linkage analysis of censored trait data

Area of Science:

• Genetics
• Biostatistics
• Quantitative Trait Analysis

Background:

• Variance-component (VC) methods are essential for mapping genes influencing quantitative traits.
• Traditional VC methods assume multivariate normality, which is violated by censored trait data.
• Censored data, arising from assay limits or medication, necessitates modified VC methods for valid linkage analysis.

Purpose of the Study:

• To introduce a novel tobit variance-component (VC) method for analyzing censored quantitative trait data.
• To compare the performance of the tobit VC method against traditional VC methods using simulation studies.
• To evaluate the tobit VC method's application to real-world censored genetic data.

Main Methods:

• Developed the tobit VC method to directly model the censoring event in quantitative trait data.
• Conducted simulation studies to assess parameter estimation bias, false-positive rates, and linkage power.
• Applied the tobit VC method to censored data from the Finland-United States Investigation of Non-Insulin-Dependent Diabetes Mellitus Genetics study.

Main Results:

• Traditional VC methods produced biased parameter estimates and increased false positives with censored data.
• The tobit VC method yielded unbiased parameter estimates and nominal type I error rates.
• The tobit VC method demonstrated a modest increase in linkage power compared to traditional methods.

Conclusions:

• The tobit VC method provides a valid and more powerful approach for linkage analysis of censored quantitative trait data.
• Traditional VC methods are unreliable for censored data, leading to significant analytical issues.
• The tobit VC method offers improved accuracy and insights in genetic studies involving censored trait measurements.