A hidden Markov tree model for testing multiple hypotheses corresponding to Gene Ontology gene sets

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Analyzing high throughput transcriptome data often involves testing predefined gene categories.
• This systematic approach generates numerous hypotheses, with logical restrictions imposed by gene category relationships.
• Existing fully Bayesian methods are powerful but computationally intensive.

Purpose of the Study:

• To develop a computationally efficient method for analyzing gene categories in transcriptome data.
• To improve upon the speed and power of existing hypothesis testing methods.
• To provide a tool that respects logical restrictions among gene categories.

Main Methods:

• Development of a hidden Markov tree model (HMTM).
• Implementation of an algorithm that is orders of magnitude faster than existing Bayesian methods.
• Validation through simulation studies and an expression quantitative trait loci (eQTL) study.

Main Results:

• The HMTM method demonstrates significantly improved computational efficiency.
• HMTM provides more powerful results compared to other methods that account for logical restrictions.
• Individual posterior probabilities of differential expression for gene sets are estimated.

Conclusions:

• The HMTM method offers a computationally efficient and powerful approach for gene category analysis.
• The method aids in the interpretation of results by providing gene set-specific probabilities.
• An R package for the HMTM method is publicly available.