OXBench

OXBench evaluates protein multiple sequence alignments to benchmark alignment algorithms and assess implications for protein structure prediction and identification of functional residues.

Key Features:

• Reference Alignments: OXBench provides reference alignments derived from protein three-dimensional structure comparisons that serve as gold-standard benchmarks for multiple sequence alignment evaluation.
• Evaluation Measures and Software: The suite includes evaluation measures and software to automatically benchmark generated alignments against these reference standards.
• Benchmark Suite: OXBench tests various alignment methods, including AMPS (a simple hierarchical algorithm) and T-COFFEE, to compare algorithm performance.
• Performance Analysis: Performance is evaluated using reference-dependent measures and independent metrics such as the STAMP Sc Score.

Scientific Applications:

• Algorithm Comparison: Comparing multiple sequence alignment algorithms to quantify relative performance across methods.
• Progress Assessment: Providing a baseline to track advancements in sequence alignment methodologies over time.
• Matrix Evaluation: Highlighting the impact of pair-score matrices on alignment accuracy, noting that improvements since 1985 have been largely attributed to enhanced matrices rather than algorithmic changes.

Methodology:

Alignments generated by various methods are evaluated against reference alignments; the suite tests AMPS using different pair-score matrices such as the BLOSUM series, demonstrating its competitive accuracy in Structurally Conserved Regions (SCRs); it compares eight different alignment algorithms (reporting T-COFFEE's superior performance on datasets with fewer than eight sequences) and computes maximum theoretical alignment accuracy and performance across a range of sequence identities.