Apollo

Apollo applies a profile hidden Markov model-based polishing algorithm to improve genome assembly accuracy by correcting sequencing errors using reads from second- and third-generation sequencing technologies.

Key Features:

• Universal compatibility: Integrates reads from both second- and third-generation sequencing technologies within a single polishing run.
• Scalability: Processes assemblies of any size without requiring segmentation of large genomes.
• Error correction mechanism: Models an assembly as a profile hidden Markov model (pHMM) and uses read-to-assembly alignment information to identify errors.
• Polishing process: Trains the pHMM with the Forward-Backward algorithm and decodes the trained model with the Viterbi algorithm to produce a polished assembly.

Scientific Applications:

• Genome assembly polishing: Produces refined assemblies with reduced sequencing errors for downstream analysis.
• Variant detection: Improves accuracy of variant calls by correcting assembly errors that can confound variant discovery.
• Functional annotation: Enhances sequence fidelity used as input for gene prediction and functional annotation pipelines.
• Comparative genomics: Provides higher-quality assemblies for comparative analyses across genomes.

Methodology:

Apollo models the assembly as a profile HMM, uses read-to-assembly alignments to train the pHMM with the Forward-Backward algorithm, and decodes the trained model with the Viterbi algorithm to generate the polished assembly.