MADOKA

MADOKA performs ultra-fast, large-scale protein structure similarity searches to identify and compare structural neighbors within the Protein Data Bank (PDB).

Key Features:

• Two-phase algorithm: Employs a novel two-phase algorithm combining a rapid pre-alignment with a precise fragment-based residue-level alignment.
• Rapid pairwise pre-alignment: Performs an initial fast alignment between pairwise structures to quickly identify potential structural neighbors.
• Fragment-based residue-level alignment: Applies a fragment-based residue-level alignment to pairs with higher preliminary similarity for precise residue-level matching.
• Database-scale PDB search: Enables searches across the Protein Data Bank (PDB), covering approximately 360,000 protein chains.
• High performance versus TM-align and SAL: Achieves approximately 6-100 times faster speeds than TM-align and SAL in massive alignments.
• Improved alignment quality: Delivers higher-quality structural alignments measured by TM-score and the number of aligned residues compared to TM-align and SAL.
• Parallel multi-core implementation: Implements parallel computation optimized for multi-core CPUs to accelerate large-scale searches.

Scientific Applications:

• Structural neighbor identification: Searching and identifying structural neighbors across the PDB for comparative studies.
• Comparative structural analysis: Quantifying structural similarity using TM-score and aligned residue counts for protein structure comparisons.
• High-throughput large-scale comparisons: Enabling massive alignment tasks for large-scale structural bioinformatics investigations.

Methodology:

MADOKA applies a two-phase algorithm consisting of an initial rapid pairwise alignment to identify candidate neighbors followed by a fragment-based residue-level alignment for pairs with higher preliminary similarity, and the implementation is parallelized for multi-core CPUs.