Spritz

Spritz predicts intrinsically disordered regions in protein sequences to identify segments that lack stable secondary or tertiary structure and to support studies of intrinsically disordered proteins (IDPs).

Key Features:

• Sequence-based prediction: Classifies ordered versus disordered regions using protein sequence data.
• Dual SVM classifiers: Employs two support vector machines to distinguish ordered and disordered protein regions.
• Length-specific modeling: SVMs are trained and benchmarked on datasets representing both long and short disordered regions.
• Benchmark evaluation: Performance was assessed in the community experiment CASP-6 for protein structure prediction.
• Improved discrimination across lengths: The dual-SVM strategy enhances prediction across varying lengths of disorder.

Scientific Applications:

• IDR annotation: Identification and annotation of intrinsically disordered regions in individual proteins and complete genome proteomes.
• Functional studies of IDPs: Support for investigations into functional roles of intrinsically disordered proteins and regions.
• Method benchmarking: Use as a predictive method for benchmarking in community evaluations such as CASP-6.

Methodology:

Spritz uses a machine learning approach with two support vector machines trained and benchmarked on sequence-based datasets representing long and short disordered regions to classify ordered versus disordered protein segments.