MetaDisorder

MetaDisorder integrates multiple disorder prediction methods to generate weighted consensus predictions of intrinsically unstructured regions (IUPs) from protein sequence data.

Key Features:

• Weighted consensus: Combines outputs from contributing predictors into a final consensus prediction weighted according to the accuracy of each method.
• Integration of diverse predictors: Aggregates results from 13 disorder predictors (arbitrarily chosen) that employ evolutionary information, energy functions, statistical analyses, and machine learning techniques.
• GSmetaDisorder3D: Implements a predictor that infers structured and unstructured regions by aligning sequences to known protein structures identified by protein fold-recognition methods.
• Meta-meta predictor (GSmetaDisorderMD): Integrates component predictors into an additional combined predictor for enhanced performance.

Scientific Applications:

• Identification of IUPs/IDRs: Predicts intrinsically unstructured proteins and intrinsically disordered regions from primary sequence information.
• Inference of structured regions: Uses fold-recognition-based alignments to infer potentially structured regions within sequences alongside disordered regions.
• Benchmarking and method development: Serves as a consensus framework evaluated in CASP competitions (CASP8, CASP9) for comparative assessment of disorder prediction performance.
• Support for structure–function studies: Provides disorder annotations useful for investigating protein structural dynamics and functional interactions.

Methodology:

Applies 13 disorder predictors to protein sequence data, generates a consensus prediction weighted by each method's accuracy, uses GSmetaDisorder3D to align sequences to known structures via protein fold-recognition methods to infer structured/unstructured regions, and combines predictors into GSmetaDisorderMD.