GPSeq

GPSeq models position-level RNA sequencing (RNA-seq) read counts with a two-parameter generalized Poisson (GP) distribution to improve estimation of gene and exon expression, normalization across samples, and detection of differential expression and splicing.

Key Features:

• Two-parameter generalized Poisson (GP) model: Models position-level read counts using a two-parameter generalized Poisson distribution to capture variability beyond the standard Poisson distribution.
• Position-level read count modeling: Operates on per-position RNA-seq read counts rather than aggregated counts to inform expression estimates.
• Improved model fit: Provides a better fit to RNA-seq count variability compared to the standard Poisson distribution.
• Gene and exon expression estimation: Produces estimates of gene and exon expression levels from RNA-seq data.
• Differential expression detection: Identifies differentially expressed genes between conditions or treatments.
• Differential splicing detection: Detects differentially spliced exons and supports analysis of alternative splicing events.
• Normalization across samples: Uses the GP model to facilitate more reasonable normalization across samples for comparative analyses.
• Sequencing bias consideration: Addresses sequencing bias and variability inherent to deep sequencing technologies in RNA-seq data modeling.

Scientific Applications:

• Transcriptome quantification: Estimating gene and exon expression levels from RNA-seq data for transcriptome characterization.
• Differential expression analysis: Identifying genes that are differentially expressed between experimental conditions or treatments.
• Differential splicing analysis: Detecting differentially spliced exons and investigating alternative splicing events.
• Comparative sample normalization: Improving normalization for comparative analyses across samples in RNA-seq studies.
• Analysis of deep sequencing datasets: Application to multiple RNA-seq datasets to provide more reliable expression and splicing inferences.

Methodology:

Implements a two-parameter generalized Poisson (GP) model for position-level RNA-seq read counts, compares model fit to the standard Poisson distribution, and uses the GP model to enable normalization across samples; applied to multiple RNA-seq datasets.