Vargas

Vargas computes heuristic-free optimal alignments of short sequencing reads to linear and graph-based genomic representations, including directed acyclic graphs (DAGs), to provide accurate gold-standard mappings for downstream genomic analyses.

Key Features:

• Heuristic-Free Algorithm: Guarantees finding the highest-scoring alignment for sequencing reads against linear and graph genomes (DAGs) without using heuristics.
• Scoring Flexibility: Supports semiglobal and local alignment modes with affine gap penalties and quality-scaled mismatch penalties, reproducing scoring functions used by Bowtie 2 and BWA-MEM.
• Graph Genome Support: Aligns reads to graph-based genomic representations, explicitly including directed acyclic graphs (DAGs).
• Computational Performance: Employs multi-core parallelization and vectorized SIMD instructions to achieve up to 456 billion cell updates per second.
• Benchmarking Capability: Provides optimal "gold standard" alignments for evaluating and improving heuristic aligners and for tuning parameters of Bowtie 2, BWA-MEM, and vg.
• Implementation: Implemented in C++.

Scientific Applications:

• Aligner Evaluation and Optimization: Serves as a reference for validating and improving heuristic alignment strategies and parameter tuning for Bowtie 2, BWA-MEM, and vg.
• Variant Calling: Supplies accurate read mappings that directly support high-confidence variant calling.
• Genome Assembly: Improves read placement accuracy relevant to assembly workflows.
• Comparative Genomics: Enables precise mapping to support comparative analyses across genomes.

Methodology:

Performs heuristic-free optimal alignment on linear and DAG-based graph genomes using semiglobal and local alignment modes with affine gap penalties and quality-scaled mismatch penalties, implemented with multi-core parallelization and SIMD vectorization in C++.