abPOA

abPOA performs partial order multiple sequence alignment by aligning sequences to directed acyclic graphs to enable efficient multiple sequence alignment, consensus calling, and long-read error correction and assembly.

Key Features:

• Partial Order Alignment to DAGs: Aligns sequences to directed acyclic graphs (DAGs) using partial order alignment to represent and merge multiple sequence relationships.
• Adaptive Banded Dynamic Programming: Employs an adaptive banded dynamic programming strategy to reduce computation while preserving alignment accuracy for large datasets.
• SIMD-optimized C implementation: Implements alignment kernels in C with Single Instruction Multiple Data (SIMD) optimizations for high-throughput performance.
• Performance vs SPOA: Demonstrates up to ~15-fold speed improvement over SPOA while maintaining comparable alignment accuracy.
• Consensus calling and MSA capability: Provides functionality for multiple sequence alignment and generation of consensus sequences from aligned inputs.

Scientific Applications:

• Long-read error correction: Used to correct errors in long-read sequencing data by producing accurate MSAs and consensus sequences.
• Genome assembly: Supports assembly workflows by providing efficient alignments and consensus information for long-read assemblies.
• Consensus sequence generation: Generates consensus sequences from aligned reads to identify common variants and representative sequences.

Methodology:

Performs partial order alignment to DAGs using adaptive banded dynamic programming implemented in a SIMD-optimized C library.