Maximum likelihood inference of imprinting and allele-specific expression from EST data

Area of Science:

• Genomics
• Transcriptomics
• Bioinformatics

Background:

• Allelic expression in diploid organisms can vary due to epigenetic factors or cis-acting polymorphisms.
• Single Nucleotide Polymorphism (SNP) allele counts from expressed sequence tags (ESTs) have been used to study genomic imprinting and allelic expression differences.
• Existing methods for analyzing allelic expression from EST data have limitations.

Purpose of the Study:

• To develop and validate a statistical framework for identifying genes with unequal representation of parental alleles in transcript data.
• To model both genomic imprinting and allele-specific expression using public transcript data.
• To improve upon existing methods for detecting allele-specific expression and imprinting.

Main Methods:

• Developed a maximum likelihood statistical model framework to analyze public transcript data.
• Applied the models to a large dataset of SNPs mapped to EST sequences.
• Utilized simulations closely matched to real data to assess model performance.
• Validated the approach using known imprinted genes and inferred candidate imprinted genes.

Main Results:

• The developed models demonstrate significantly improved performance over existing methods for analyzing allelic expression.
• Evidence of undiscovered imprinted genes was found in known imprinted regions.
• Over half of the genes with substantial available data exhibit some form of allele-specific expression.
• A set of candidate imprinted genes was inferred, some located near known imprinted genes.

Conclusions:

• The new statistical models provide a powerful and efficient method for detecting allele-specific expression and genomic imprinting.
• The study suggests that allele-specific expression is a common phenomenon in diploid organisms.
• Further investigation into candidate imprinted genes and regions is warranted.