BEXCIS

BEXCIS estimates the degree of skewness in X chromosome inactivation (XCI) to quantify allele-specific inactivation patterns relevant to dosage compensation and analyses of X-linked diseases.

Key Features:

• Bayesian models (BN and BU): Implements two Bayesian models—a BN model with a normal prior and a BU model with a uniform prior—using prior information about the expected XCI skewness range (0 to 2).
• Penalized Fieller's method: Implements a penalized Fieller's method to produce penalized point estimates of XCI skewness and corresponding confidence intervals.
• Statistical issue mitigation: Methods are designed to mitigate extreme point estimates, noninformative intervals, empty sets, and discontinuous intervals encountered in traditional analyses.
• Simulation-based evaluation: Uses simulation studies to assess estimator performance using metrics such as mean squared error, coverage probability, and interval width.
• BN performance advantage: Simulation results indicate the BN method attains the lowest mean squared error for point estimation while controlling coverage probability and minimizing median and variation of interval width.

Scientific Applications:

• Dosage compensation and X-linked disease analysis: Quantifies XCI skewness for studies of dosage compensation and associations with X-linked diseases.
• Graves' disease data: Applied to Graves' disease data and identified SNP rs3827440 with skewness toward the C allele.
• Minnesota Center for Twin and Family Research data: Applied to data from the Minnesota Center for Twin and Family Research to detect potential skewed XCI events.

Methodology:

BEXCIS applies two Bayesian models (BN with a normal prior and BU with a uniform prior) constrained to the expected XCI skewness range (0–2), implements a penalized Fieller's method for penalized point estimation and confidence intervals, and evaluates performance by simulation measuring mean squared error, coverage probability, and interval width.