BM-BC

BM-BC improves base calling accuracy in Solexa-GA sequencing by applying a hierarchical Bayesian model to position-specific intensity measurements of nucleotides A, C, G, and T.

Key Features:

• Hierarchical Bayesian model: Applies a hierarchical Bayesian statistical framework for inferring true bases from signal intensities.
• Noise modeling: Explicitly models fading, phasing, and cross-talk between channels within the inference framework.
• Position-specific intensity analysis: Analyzes position-specific intensity measurements corresponding to A, C, G, and T at each genomic location.
• Per-base probability scores: Computes a probability score for each base call that quantifies confidence in nucleotide identification.
• False discovery rate estimation: Uses probability scores to estimate the false discovery rate (FDR) of base calls.
• Sequence ranking by precision: Ranks sequences by precision to support downstream analyses such as sequence alignment.
• Short-read applicability: Targets base calling for short DNA and RNA reads generated by Solexa-GA sequencing.
• Improved precision: Enhances base calling precision relative to existing methods.

Scientific Applications:

• Base calling for short-read sequencing: Produces more accurate base calls for short DNA and RNA reads from Solexa-GA sequencing.
• Variant detection: Supports higher-confidence variant detection through probability-weighted base calls.
• Transcriptome analysis: Enables more reliable transcriptome analyses by providing confidence-scored base calls.
• Sequence alignment: Facilitates sequence alignment by ranking reads based on call precision.
• Error assessment and quality control: Allows assessment and management of base-call errors via FDR estimation.

Methodology:

Analyzes position-specific intensity measurements and implements a hierarchical Bayesian model that accounts for fading, phasing, and cross-talk; computes per-base probability scores which are used to estimate FDR and rank sequences.