Where differences resemble: sequence-feature analysis in curated databases of intrinsically disordered proteins

Area of Science:

• * Bioinformatics
• * Structural Biology
• * Computational Biology

Background:

• * Intrinsic disorder (ID) in proteins plays a critical role in cellular interactions.
• * Growing recognition of ID's importance necessitates detailed identification and characterization.
• * The existence of different 'flavors' or sub-phenomena of ID remains an open question.

Purpose of the Study:

• * To compare manually curated, experimentally validated ID collections.
• * To assess sequence-based features differentiating various ID databases.
• * To understand the representation of different ID phenomena across databases.

Main Methods:

• * Analysis of sequence-based features: amino acid composition, length, taxa, and Gene Ontology terms.
• * Comparison of content from multiple ID databases: DIBS, DisProt, IDEAL, MFIB, FuzDB, ELM, and UniProt.
• * Statistical assessment of feature differences across datasets.

Main Results:

• * Most sequence-based features showed no significant statistical differences across ID databases.
• * ELM database exhibited statistically different features compared to others.
• * FuzDB shares approximately half of its entries with DisProt.
• * Generally, different ID databases describe similar intrinsic disorder phenomena.

Conclusions:

• * DisProt, the largest database, offers a comprehensive representation of ID diversity.
• * The majority of analyzed features do not distinguish between ID 'flavors' across databases.
• * Findings suggest a general similarity in phenomena described by various ID databases.