MBttest

MBttest applies beta and binomial distribution-based statistical modeling to NGS sequence count data to detect differential expression and genome-wide differential splicing while providing high power and conservative false discovery rate estimation for small-sample transcriptomic analyses.

Key Features:

• Applicability to count data: Tailored for sequence count data generated by next-generation sequencing (NGS) rather than microarray measurements.
• High power and conservative FDR estimation: Maintains high statistical power while conservatively estimating false discovery rate (FDR) for differential expression calls.
• Stability with limited replicates: Provides stable results when analyzing count data from small sample sizes or limited replicate libraries.
• Robust performance across datasets: Demonstrated superior performance on both simulated and real-world transcriptomic datasets compared with existing statistical methods.
• Experimental validation and splicing extension: Differential expression results have been validated by quantitative PCR (qPCR), and the method can be extended to genome-wide detection of differential splicing events.

Scientific Applications:

• Differential expression analysis: Detection of differentially expressed genes or mRNA isoforms from NGS count data, including studies with rare tissues or specific developmental stages and other small-sample scenarios.
• Alternative splicing analysis: Genome-wide detection of differential splicing events using count-based transcriptomic data.

Methodology:

Implements a statistical framework based on beta and binomial distributions to model discrete NGS count data, account for variability in small samples, and estimate FDR for differential expression and splicing detection.