Bayesembler

Bayesembler reconstructs full-length transcripts and quantifies their abundances from RNA sequencing (RNA-seq) data using a Bayesian probabilistic model.

Key Features:

• Bayesian Model: A comprehensive Bayesian model explicitly captures the RNA-seq data generation process to enable probabilistic inference of transcript composition and abundance.
• Gibbs Sampling: Gibbs sampling is used to generate samples from the posterior distribution over possible transcripts and their abundance values.
• Posterior Distribution Analysis: Transcript assemblies are selected by analyzing frequencies of transcripts observed in the posterior samples to identify the most probable configurations.
• Performance Improvements: Empirical evaluations on simulated and real datasets report increased sensitivity and precision relative to other state-of-the-art assemblers.

Scientific Applications:

• Gene Expression Studies: Reconstruction and quantification of transcripts from RNA-seq data supports accurate gene expression profiling.
• Alternative Splicing Analysis: Identification of transcript variants from posterior samples facilitates detection and analysis of alternative splicing events.
• Comparative Transcriptomics: Reliable transcript assemblies enable comparative analyses across conditions or species to study regulatory and evolutionary differences.

Methodology:

Bayesembler applies a Bayesian framework that models the RNA-seq generation process, uses Gibbs sampling to explore the posterior distribution over transcripts and abundances, and selects assemblies based on posterior sample frequencies.