DisoFLAG

DisoFLAG predicts protein intrinsic disorder and associated functions from amino acid sequences to characterize intrinsically disordered proteins and regions (IDPs/IDRs).

Key Features:

• Graph-Based Interaction Protein Language Model (GiPLM): DisoFLAG integrates semantic information from pre-trained protein language models into graph-based interaction units to enhance correlation of semantic representations across multiple disordered functions.
• Comprehensive Function Prediction: DisoFLAG predicts six disordered functions—Protein-binding, DNA-binding, RNA-binding, Ion-binding, Lipid-binding, and Flexible linker—using sequence information alone.
• Sequence-only input: DisoFLAG requires only amino acid sequences as input to generate disorder and function predictions.
• High Predictive Accuracy: DisoFLAG's performance was evaluated in Critical Assessment of protein Intrinsic Disorder (CAID) experiments and demonstrated accurate and comprehensive predictions.

Scientific Applications:

• IDP/IDR functional annotation: Predicting intrinsic disorder and the six associated functions to aid characterization of intrinsically disordered proteins and regions in molecular studies.
• Protein interaction and mechanism inference: Informing studies of protein interactions and cellular mechanisms by identifying disorder-mediated binding and flexible linker regions.

Methodology:

DisoFLAG takes amino acid sequences as input and leverages the GiPLM framework, integrating semantic representations from pre-trained protein language models into graph-based interaction units to predict intrinsic disorder and associated functions.