SHARK.webservice

SHARK.webservice performs alignment-free homology classification of intrinsically disordered regions (IDRs) by computing k-mer-based SHARK-scores and applying the SHARK-dive machine-learning classifier to detect evolutionary homology and functional analogy.

Key Features:

• Alignment-free k-mer comparison: Computes k-mer matches and assesses overall amino acid composition to identify shared short regions between sequences, yielding SHARK-scores.
• Machine learning classifier (SHARK-dive): Uses a classifier trained on disordered and difficult-to-align protein regions to detect evolutionary homology and functional analogy from SHARK-scores.
• Enhanced sensitivity versus BLAST and HMMER: Provides improved detection of low sequence identity homologs in intrinsically disordered sequences compared to alignment-based methods such as BLAST and HMMER.
• Databases and precomputed IDR-omes: Supports comparisons against extensive sequence databases and includes precomputed sets of IDR sequences from 16 model organism proteomes for species-targeted searches.

Scientific Applications:

• IDP homology detection: Identification of evolutionary relationships among intrinsically disordered proteins and regions that are unalignable by conventional methods.
• Sequence–function relationship analysis: Investigation of links between disordered region sequence features and functional analogy.
• Evolutionary biology: Detection of distant homologs and conserved features in low sequence identity IDRs for evolutionary studies.
• Functional genomics: Targeted searches within species-specific IDR-omes to support genome-scale analyses of disordered regions.

Methodology:

Computes k-mer-based comparisons and overall amino acid composition to produce SHARK-scores, then classifies sequences with the SHARK-dive machine-learning model trained on disordered/difficult-to-align regions and enables comparison against databases including precomputed IDR sets from 16 model organism proteomes.