Selection of X-chromosome Inactivation Model

Area of Science:

• Genetics
• Statistical Genetics
• Computational Biology

Background:

• X-chromosome inactivation (XCI) is a complex process with random, skewed, and escaping patterns.
• Previous association tests for X-chromosomal single-nucleotide polymorphisms (SNPs) did not formally distinguish between these XCI models.

Purpose of the Study:

• To develop and evaluate a new statistical approach for selecting the most likely XCI model.
• To compare the performance of the proposed method against an existing approach using simulations and real data.

Main Methods:

• Proposed a likelihood ratio comparison (comp-LLR) procedure, inspired by the Cox test, to formally compare different XCI models.
• Conducted simulation studies to assess the accuracy of the comp-LLR approach versus the max-LLR approach.
• Applied both methods to a head and neck cancer genetic study.

Main Results:

• The comp-LLR approach demonstrated a higher probability of correctly identifying the underlying XCI model compared to the max-LLR approach.
• The comp-LLR method showed superior performance for random XCI, escaping XCI, and skewed XCI towards a deleterious allele.
• Analysis of head and neck cancer data provided insights into XCI processes for X-chromosomal variants.

Conclusions:

• The comp-LLR approach offers a statistically robust method for selecting the most appropriate XCI model.
• This advancement can improve the accuracy of genetic association studies involving X-chromosomal variants.
• The findings have implications for understanding the genetic basis of diseases influenced by XCI patterns.