A Bayesian approach for estimating allele-specific expression from RNA-Seq data with diploid genomes

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• RNA-sequencing (RNA-Seq) is widely used for transcriptome profiling.
• Accurate allele-specific expression (ASE) estimation is challenging due to reference genome limitations and high sequence similarity between alleles.
• Existing methods struggle with precise read alignment to correct alleles in diploid genomes.

Purpose of the Study:

• To develop a novel Bayesian approach for accurate ASE estimation from RNA-Seq data.
• To improve the quantification of isoform expression levels.
• To identify allele-specific expression patterns and chromosomal inactivation.

Main Methods:

• A Bayesian statistical framework was developed to model haploid choice as a hidden variable.
• Variational Bayesian inference was employed for simultaneous estimation of hidden variables and isoform expression levels.
• The proposed method, ASE-TIGAR, was validated using simulation data and real RNA-Seq data from human lymphoblastoid cell line GM12878.

Main Results:

• The proposed Bayesian approach demonstrated superior performance in identifying ASE compared to existing methods.
• ASE-TIGAR achieved more accurate quantification of isoform expression levels than TIGAR2, RSEM, and Cufflinks.
• Analysis of GM12878 data revealed novel autosomal ASE genes and identified skewed paternal X-chromosome inactivation.

Conclusions:

• ASE-TIGAR provides accurate allele-specific gene expression estimation from RNA-Seq data.
• The study highlights the value of incorporating personal genomic information for precise ASE analysis.
• The ASE-TIGAR method is publicly available for research use.