MMBGX: a method for estimating expression at the isoform level and detecting differential splicing using...

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Whole-transcript microarrays, such as Affymetrix Gene and Exon arrays, interrogate gene expression at the exon level.
• Probes on these arrays can hybridize to multiple genomic locations or alternative transcript isoforms, leading to inaccurate expression estimates with current methods.
• Existing statistical approaches do not account for multi-mapping probes, resulting in inflated gene expression levels.

Purpose of the Study:

• To develop a novel statistical method, Multi-Mapping Bayesian Gene eXpression (MMBGX), to address the challenge of multi-mapping probes in whole-transcript array data.
• To improve the accuracy of gene-level expression estimates by disaggregating signals from multi-match probes.
• To enable the estimation of expression levels for individual splice variants and detect differential splicing events.

Main Methods:

• Development and application of the Multi-Mapping Bayesian Gene eXpression (MMBGX) algorithm.
• Disaggregation of signals from probes that match multiple transcripts.
• Validation using simulated data, a tissue mixture dataset, and RT-PCR confirmation of alternative isoform expression.
• Comparison with standard exon-level approaches for differential splicing detection.

Main Results:

• MMBGX effectively removes the upward bias in gene-level expression estimates derived from Gene arrays.
• MMBGX enables the disaggregation of signals for alternative transcripts from Exon arrays, providing expression estimates for individual splice variants.
• The method accurately estimates alternative isoform expression within a single experimental condition, validated by RT-PCR.
• MMBGX demonstrates a lower error rate in detecting differential splicing compared to standard methods on a colon cancer dataset.

Conclusions:

• MMBGX provides a robust solution for accurate gene expression quantification from whole-transcript arrays, particularly when dealing with multi-mapping probes.
• The method enhances the analysis of alternative splicing by enabling the quantification of individual splice variants.
• MMBGX offers improved accuracy and reduced error rates for differential splicing detection, with significant implications for cancer research and other applications.