SABmark--a benchmark for sequence alignment that covers the entire known fold space

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• Accurate protein sequence alignment is crucial for understanding protein function and evolution.
• Existing benchmarks may not adequately represent the full spectrum of sequence similarity encountered in real-world datasets.
• The Structural Classification of Proteins (SCOP) database provides a hierarchical classification of protein folds.

Purpose of the Study:

• To introduce the Sequence Alignment Benchmark (SABmark) as a comprehensive resource for evaluating sequence alignment algorithms.
• To provide benchmark sets covering a wide range of sequence similarities, including challenging low-similarity regions.
• To assess the performance of alignment tools on datasets containing potentially misleading, unalignable sequences.

Main Methods:

• Generation of multiple sequence alignment problems derived from the SCOP classification.
• Creation of two main benchmark sets: Twilight Zone (very low to low similarity) and Superfamilies (low to intermediate similarity).
• Development of alternate versions of these sets by incorporating unalignable but superficially similar sequences.

Main Results:

• SABmark encompasses the entire known protein fold space.
• The benchmark sets offer varying degrees of sequence similarity, from very low to intermediate.
• Alternate sets introduce complexity by including sequences that are difficult to align correctly.

Conclusions:

• SABmark provides a robust and diverse platform for benchmarking protein sequence alignment methods.
• The inclusion of challenging sequences enhances the evaluation of alignment tool sensitivity and specificity.
• This benchmark facilitates the development and refinement of more accurate sequence alignment algorithms.