SHARK.dive

SHARK.dive evaluates evolutionary homology and functional analogy among intrinsically disordered regions (IDRs) and other unalignable protein segments using an alignment-free, machine learning-based SHARK (Similarity/Homology Assessment by Relating K-mers) algorithm.

Key Features:

• Alignment-Free Homology Assessment: Compares sequences without relying on traditional alignments, making it suitable for IDRs and structurally flexible regions.
• Machine learning-based homology classifier: Uses a classifier trained on disordered and challenging-to-align sequences to detect remote homology.
• Detection of Functional Analogy: Identifies functionally analogous IDRs that are sequence-dissimilar but share functional properties.
• Proteome-Wide Predictions: Enables large-scale, proteome-level prediction and annotation of unalignable protein regions.
• Identification of Cryptic Sequence Properties: Detects hidden sequence motifs and properties that contribute to remote homology and functional analogy and yields interpretable hypotheses.
• Experimental Verification: Produces experimentally verifiable insights into the evolutionary and functional aspects of IDRs.

Scientific Applications:

• Homology detection in IDRs: Assess evolutionary relationships among intrinsically disordered and unalignable protein segments.
• Functional annotation of disordered regions: Infer functional roles of IDRs, including regulatory functions and participation in biomolecular condensates.
• Proteome-scale annotation: Systematically analyze and annotate unalignable regions across entire proteomes.
• Hypothesis generation for experiments: Provide testable hypotheses about sequence determinants of homology and function for experimental validation.

Methodology:

Alignment-free sequence comparison using SHARK (Similarity/Homology Assessment by Relating K-mers) combined with a machine learning homology classifier trained on disordered and challenging-to-align sequences.