MMBGX

MMBGX applies a Bayesian model to assign whole-transcript Affymetrix GeneChip probe signals to specific transcripts and isoforms, improving isoform-level expression estimation and differential splicing detection.

Key Features:

• Probe Signal Disaggregation: Disaggregates signals from probes that hybridize to multiple transcripts or alternative isoforms (multi-match probes) to refine expression estimates.
• Bias Correction at Gene Level: Reduces upward bias in gene-level expression estimates on Gene arrays by attributing probe signals to their respective transcripts.
• Isoform-Specific Expression Estimates: Separates signal contributions among alternative transcripts on Exon arrays to estimate individual splice-variant expression.
• Differential Splicing Detection: Detects differential splicing events with lower error rates than standard exon-level approaches by leveraging isoform-resolved signal assignment.

Scientific Applications:

• Expression Analysis: Generates isoform-level expression profiles for studies of gene regulation, transcript diversity, and functional genomics using GeneChips, Gene arrays, and Exon arrays.
• Disease Research: Facilitates identification of splicing alterations associated with disease, with demonstrated improved accuracy on a colon cancer dataset compared to conventional methods.

Methodology:

MMBGX employs a Bayesian framework that models the distribution of probe signals across multiple potential targets and computes probabilistic assignments (posterior likelihoods) of probe-target interactions for isoform-level expression estimation.