A mixture distribution approach for assessing genetic impact from twin study

Area of Science:

• Quantitative genetics
• Biostatistics
• Human genetics

Background:

• Disentangling genetic from environmental influences on biological traits is a persistent challenge in scientific research.
• Twin studies are a common approach, but conventional correlation methods are unsuitable for unordered twin data due to unknown genetic ordering.

Purpose of the Study:

• To develop a robust statistical method for estimating genetic impacts on biological features using twin study data.
• To address the limitations of existing methods in handling unordered twin data and improve the accuracy of genetic correlation estimation.

Main Methods:

• A novel approach modeling twin data using a mixture bivariate distribution with two likelihood functions: separate and combined estimation for monozygotic and dizygotic twins.
• The combined likelihood function is proposed to overcome slow convergence issues inherent in mixture distribution estimation.

Main Results:

• Both separate and combined likelihood estimators are demonstrated to be statistically consistent.
• The combined likelihood approach achieves root-n consistency for correlation coefficient estimation, enabling more effective statistical inference.
• The method's efficacy is validated through its application in a twin study focusing on immune traits.

Conclusions:

• The developed statistical framework provides a consistent and efficient method for analyzing unordered twin data.
• This approach enhances the ability to accurately assess the collective genetic impact on biological traits, as exemplified by immune system variations.