FBB

FBB computes a Bayesian lower bound on the probability that an RNA-seq read alignment is correct by integrating per-base quality scores from SAM-formatted aligners.

Key Features:

• Probabilistic Integration: Calculates a lower bound on the probability of correct alignment for any fast RNA-seq aligner that generates SAM files by leveraging read quality scores.
• Bayesian Bound Calculation: Implements two theorems that enable efficient computation of the Bayesian bound and that under specific conditions yield exact equality.
• Comparative Analysis: Maps aligner command options to standardized FBB reference values to enable comparison across different aligners using a consistent metric.
• Strand-Specific Evaluation: Performs error analysis for stranded paired read RNA-seq data, incorporating pair distance and strand-specific alignment error metrics.
• Broad Applicability: Has been evaluated with multiple RNA-seq aligners, including Bowtie, Bowtie 2, SHRiMP2, Soap 2, and Novoalign.

Scientific Applications:

• Alignment Quality Assessment: Provides a probabilistic measure to assess the likelihood that individual alignments are correct using per-base quality scores.
• Aligner Comparison and Calibration: Supplies a standardized reference for comparing and calibrating different RNA-seq aligners and their command options.
• Strand- and Distance-Specific Error Analysis: Enables analysis of strand-specific alignment errors and error dependence on paired-read distances in stranded paired read RNA-seq data.

Methodology:

Computes a lower bound on alignment correctness probability using per-base quality scores from SAM files; applies two theorems to efficiently compute the Bayesian bound (with exact equality under specified conditions); maps aligner command options to FBB reference values for comparative analysis; analyzes stranded paired read RNA-seq data using pair-distance and strand-specific error metrics.