Generalized F accelerated failure time model for mapping survival trait loci

Area of Science:

• Genetics
• Biostatistics
• Quantitative Trait Loci (QTL) analysis

Background:

• Survival analysis models like Cox proportional hazards model (PHM) and accelerated failure time (AFT) models are crucial in biological research.
• AFT models offer greater flexibility in fitting survival data compared to PHM.
• Identifying genetic factors influencing survival traits is essential for understanding disease mechanisms.

Purpose of the Study:

• To develop a general parametric accelerated failure time (AFT) model for identifying survival trait loci.
• To incorporate a flexible generalized F distribution as the baseline survival distribution within the AFT model.
• To evaluate the model's performance using simulations and a real-world dataset of mouse survival.

Main Methods:

• Development of a general parametric AFT model.
• Specification of the generalized F distribution as the baseline survival distribution.
• Application of the Expectation-Maximization (EM) algorithm for maximum likelihood estimation.
• Conducting simulation studies to assess model flexibility and utility.
• Analysis of survival time data in mice with hyperoxic acute lung injury (HALI).

Main Results:

• The proposed AFT model with a generalized F distribution demonstrated flexibility and utility in simulations.
• The generalized F distribution outperformed six other survival distributions in analyzing HALI survival data.
• Four quantitative trait loci (QTLs) controlling differential HALI survival were identified in an F(2) mouse population.

Conclusions:

• The developed parametric AFT model provides a robust framework for survival trait locus identification.
• The generalized F distribution is a suitable and flexible choice for modeling complex survival data.
• This approach successfully identified genetic factors influencing survival in a mouse model of lung injury.