About DiSCo

DiSCo (Disordered Structural Complexes) is a curated database of protein complexes derived from the Protein Data Bank (PDB) in which at least one protein chain is predicted to be intrinsically disordered. The database bridges the gap between experimental structural data and the functional study of intrinsically disordered proteins (IDPs), offering unique insights into disorder-mediated interactions.

Why DiSCo?

Intrinsic disorder plays a crucial role in cellular regulation, signaling, and complex formation, yet remains notoriously difficult to characterize experimentally. Disordered regions are often missing or poorly resolved in high-resolution structures, leading to an underrepresentation of IDPs in structural databases. With the rise of AlphaFold2 and other predictive tools, it's now possible to systematically identify disordered regions, even in proteins with known structures.

DiSCo addresses this challenge by:

• Identifying PDB complexes where one or more chains are predicted to be disordered based on AlphaFold2 pLDDT scores.
• Enriching these predictions with complementary disorder annotations from AIUPred and AIUPred-binding, based on biophysically grounded principles.
• Providing a functionally annotated, structure-integrated view of disorder in the context of experimentally solved complexes.

What DiSCo Provides

Each DiSCo entry corresponds to a non-redundant structural complex involving at least one disordered partner. For every entry, we provide:

• Mapped UniProt IDs for all chains.
• Disorder predictions from AlphaFold2, AIUPred, and AIUPred-binding.
• Domain annotations for structured interaction partners using Pfam, InterPro, and Gene3D.
• Overlapping Gene Ontology (GO) terms between interacting partners, highlighting shared functional roles.
• 3D visualization via Mol* of the PDB structure with highlighted disordered regions.
• Downloadable AI-ready datasets for machine learning and benchmarking.

Who is DiSCo for?

DiSCo is designed for:

• Researchers studying protein disorder, phase separation, or disorder-mediated interactions.
• Developers of machine learning models for disorder or interface prediction.
• Structural biologists looking for functional insights into complexes involving IDPs.

How to Use DiSCo

• Browse or search for complexes by PDB ID, UniProt accession
• Filter entries based on disorder score thresholds or domain types.
• Visualize disorder and interaction interfaces directly in the browser.
• Download curated subsets or the entire database for offline analysis or model training.