Testing the hypothesis of tissue selectivity: the intersection-union test and a Bayesian approach.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Standard statistical tests are inadequate for identifying genes with preferential expression in specific tissues or conditions.
• The intersection-union test (IUT) is a known method but is conservative and limited in the information it provides.
• The IUT only considers the least differing tissue pairs, potentially missing broader expression patterns.

Purpose of the Study:

• To develop a novel Bayesian procedure for quantifying evidence of selective gene over-expression.
• To compare the performance of the proposed Bayesian method against the IUT in identifying selectively expressed genes.
• To demonstrate the utility of the Bayesian method in analyzing real-world gene expression data.

Main Methods:

• A Bayesian statistical procedure was developed to assess selective over-expression based on overall expression profiles.
• A simulation study was conducted to evaluate the proposed method's performance against the IUT.
• The Bayesian method was applied to publicly available gene expression data from 22 tissues.

Main Results:

• The Bayesian procedure demonstrated superior performance compared to the IUT in identifying selectively expressed genes.
• Application to real data revealed that the Bayesian method successfully selected genes with functions relevant to specific tissue characteristics.
• The method is also capable of identifying underexpressed genes within a particular tissue.

Conclusions:

• The proposed Bayesian method offers a more sensitive and informative approach for detecting selective gene over-expression than the IUT.
• This method provides valuable insights into tissue-specific gene functions and expression patterns.
• Software for implementing both the IUT and the Bayesian procedure is available for MATLAB and R.