Skip to content
Frontiers
Search

Mini Review

Frontiers in Biophysics ·

Volume 4 - 2026 | https://doi.org/10.3389/frbis.2026.1880887

Intrinsically disordered regions as drivers of protein aggregation: mechanisms, phase separation, and emerging predictive frameworks

  • Rahul Kaushik1*corresponding author
  • Suyong Re1
  1. Laboratory of In-silico Design, Artificial Intelligence Center for Health and Biomedical Research, National Institutes of Biomedical Innovation, Health and Nutrition (NIBN), Settsu, Osaka, Japan
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Intrinsically disordered proteins (IDPs) and intrinsically disordered regions (IDRs) lack stable tertiary structures yet perform essential roles in cellular signaling, molecular recognition, transcriptional regulation, and biomolecular assembly. Their conformational flexibility enables functional adaptability but also increases susceptibility to aberrant intermolecular interactions and protein aggregation. Unlike folded proteins, aggregation in IDPs arises from transient conformational ensembles that expose cryptic aggregation-prone regions (APRs), facilitating oligomerization and fibril formation under specific cellular and environmental conditions. Several studies have further established a mechanistic relationship between intrinsic disorder, liquid–liquid phase separation (LLPS), and pathological aggregation, where dynamic condensates can undergo maturation into irreversible amyloid-like assemblies. These transitions are strongly influenced by sequence grammar, charge distribution, aromatic residue patterning, post-translational modifications, molecular crowding, and proteostasis regulation. This mini-review summarizes the molecular principles governing aggregation in disordered systems, with emphasis on conformational ensemble dynamics, disorder-to-order transitions, and the interplay between LLPS and fibrillization. The review further discusses computational approaches used to predict aggregation propensity in IDRs, including classical physicochemical predictors, ensemble-aware simulations, molecular dynamics frameworks, and emerging protein language model-based methods. Further, integration of artificial intelligence, structural biophysics, and multiscale modeling have substantially improved understanding of disorder-driven aggregation pathways. Collectively, these findings support a unified framework in which sequence composition, conformational heterogeneity, and cellular environment cooperatively regulate functional assembly and pathological aggregation in intrinsically disordered proteins.

Keywords

1 Introduction

Intrinsically disordered regions (IDRs) and intrinsically disordered proteins (IDPs) have emerged as central components of cellular organization, signaling, and molecular recognition. Unlike folded globular proteins, IDRs lack stable tertiary structure under physiological conditions and instead populate dynamic conformational ensembles that enable functional flexibility and interaction promiscuity (Csizmok et al., 2016; Uversky, 2021; Lotthammer et al., 2024; Kumar et al., 2021a). These disordered regions are highly enriched in eukaryotic regulatory proteins, signaling hubs, viral proteomes, and RNA-binding proteins, where structural plasticity facilitates context-dependent interactions and rapid adaptive responses (Uversky, 2021; Lotthammer et al., 2024; Kumar et al., 2021a; Santos et al., 2021). Increasing evidence suggests that intrinsic disorder is not merely an exception to the classical sequence–structure–function paradigm but rather represents an evolutionarily conserved organizational principle governing diverse biological processes (Csizmok et al., 2016; Uversky, 2021).

The physicochemical properties of IDRs, including low hydrophobicity, high net charge, low sequence complexity, and enrichment in polar residues, prevent stable hydrophobic collapse while promoting extensive conformational sampling (Uversky, 2021; Lotthammer et al., 2024). However, the same sequence features that support functional adaptability also create susceptibility to aberrant intermolecular interactions and protein aggregation (Santos et al., 2021; Strodel, 2021). Unlike folded proteins, where aggregation is generally initiated by partial unfolding and exposure of buried hydrophobic cores, aggregation in IDRs arises from transient ensemble states that expose cryptic aggregation-prone regions (APRs) (Santos et al., 2021; Strodel, 2021). These rare conformations can nucleate oligomerization and fibril formation, particularly under conditions that alter electrostatic balance, local concentration, or intermolecular interactions.

Several studies have established a close mechanistic relationship between intrinsic disorder, liquid–liquid phase separation (LLPS), and pathological aggregation (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Visser et al., 2024). IDRs frequently drive LLPS through weak multivalent interactions involving charged, aromatic, and low-complexity sequence motifs (Banani et al., 2017; Bremer et al., 2022; Alberti et al., 2025). Although biomolecular condensates formed through LLPS are generally dynamic and reversible, prolonged residence within dense phases can promote structural maturation into gel-like or amyloid assemblies (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Visser et al., 2024). This disorder–phase separation–aggregation continuum is particularly relevant to neurodegenerative diseases involving α-synuclein, tau, FUS, and TDP-43, where functional condensates progressively transition into irreversible fibrillar aggregates (Ahmad et al., 2022; Uversky, 2007; Utami et al., 2025; Ishiguro et al., 2021; Hayes and Kalab, 2022). The balance between reversible assembly and pathological aggregation is strongly influenced by sequence grammar, post-translational modifications, molecular crowding, and cellular stress conditions (Vendruscolo and Fuxreiter, 2022; Bremer et al., 2022; Uversky, 2007).

Understanding aggregation in disordered systems remains challenging because classical aggregation predictors were primarily developed for folded proteins and rely heavily on hydrophobicity and β-sheet propensity (Fernandez-Escamilla et al., 2004; Walsh et al., 2014; Tartaglia and Vendruscolo, 2008; Zambrano et al., 2015). Such approaches often fail to capture ensemble heterogeneity, sequence patterning, and context-dependent masking of APRs within IDRs (Santos et al., 2021; Zambrano et al., 2015). Emerging computational approaches increasingly integrate disorder propensity, charge distribution, sequence complexity, and machine learning-based approaches to improve aggregation prediction in intrinsically disordered proteins. Emerging protein language models (PLMs), ensemble-aware frameworks, molecular dynamics (MD) simulations, cryo-electron microscopy, and integrative artificial intelligence approaches are collectively reshaping current understanding of disorder-driven aggregation pathways by capturing long-range sequence dependencies and conformational heterogeneity (Lotthammer et al., 2024; Ginell et al., 2025; Novak et al., 2026; Kaushik and Launey, 2024; Mall et al., 2025; Yagi et al., 2022; Ghosh et al., 2024; Hollingsworth and Dror, 2018). This review presents a multiscale perspective in which sequence grammar, conformational landscapes, phase behavior, and cellular environment collectively govern aggregation outcomes. In particular, the review highlights the mechanistic relationship between intrinsic disorder, LLPS, and fibrillization, along with evolving methodologies in computational prediction frameworks relevant to disease biology, protein engineering, and therapeutic intervention (Figure 1).

A depiction of continuum of intrinsic disorder, phase separation, and aggregation in intrinsically disordered proteins. Sequence-encoded features of intrinsically disordered regions (IDRs) promote formation of dynamic conformational ensembles that continuously sample expanded and partially compact states. Environmental and cellular modulators regulate intermolecular interactions and conformational transitions. Dynamic condensates formed through weak multivalent interactions may progressively mature into oligomeric and amyloid-like assemblies characterized by increased β-sheet organization and reduced reversibility. Disease-associated proteins including α-synuclein, tau, FUS, and TDP-43 exemplify this phase separation–aggregation continuum observed in neurodegenerative disorders. The lower panel summarizes representative computational approaches used to investigate disorder-driven aggregation prediction frameworks. Together, these integrative approaches provide mechanistic insight into the relationship between intrinsic disorder, condensate dynamics, and pathological aggregation. (created with [BioRender.com](http://BioRender.com))
Figure 1. A depiction of continuum of intrinsic disorder, phase separation, and aggregation in intrinsically disordered proteins. Sequence-encoded features of intrinsically disordered regions (IDRs) promote formation of dynamic conformational ensembles that continuously sample expanded and partially compact states. Environmental and cellular modulators regulate intermolecular interactions and conformational transitions. Dynamic condensates formed through weak multivalent interactions may progressively mature into oligomeric and amyloid-like assemblies characterized by increased β-sheet organization and reduced reversibility. Disease-associated proteins including α-synuclein, tau, FUS, and TDP-43 exemplify this phase separation–aggregation continuum observed in neurodegenerative disorders. The lower panel summarizes representative computational approaches used to investigate disorder-driven aggregation prediction frameworks. Together, these integrative approaches provide mechanistic insight into the relationship between intrinsic disorder, condensate dynamics, and pathological aggregation. (created with BioRender.com)

2 Conformational landscapes of intrinsically disordered regions

Intrinsic disorder arises from sequence-encoded physicochemical features that prevent the formation of stable tertiary structures while enabling extensive conformational sampling and dynamic intermolecular interactions (Csizmok et al., 2016; Uversky, 2021; Lotthammer et al., 2024). IDRs are typically enriched in polar and charged residues, including glutamate, lysine, arginine, glutamine, and serine, while being depleted in bulky hydrophobic amino acids required for stable hydrophobic core formation (Uversky, 2021; Lotthammer et al., 2024). These compositional biases generate flexible conformational ensembles characterized by low sequence complexity, high solvent accessibility, and elevated net charge per residue (NCPR) (Uversky, 2021; Strodel, 2021). Key physicochemical features distinguishing IDRs from folded protein domains and their associated biophysical consequences are summarized in Table 1. Comparative analyses of viral and eukaryotic proteomes further demonstrate that intrinsic disorder is selectively enriched in proteins involved in signaling, transcriptional regulation, host-pathogen interactions, and stress-responsive pathways (Kumar et al., 2021a; Kumar et al., 2021b). Recent large-scale analyses of bat proteomes further demonstrated extensive enrichment of intrinsic disorder in proteins associated with host adaptation, immune modulation, and viral interaction networks (Kumar et al., 2025).

TABLE 1. Physicochemical properties and compositional features distinguishing intrinsically disordered regions (IDRs) from folded protein domains. Intrinsically disordered regions differ substantially from folded protein domains in sequence composition, charge distribution, hydrophobicity, and structural organization.

PropertyFolded/Globular domainsIntrinsically disordered regionsBiophysical consequence
Net charge per residueLow to moderate; balanced distribution of charged residuesHigh; enriched in charged residues with distinct charge patterningPromotes electrostatic repulsion, expanded conformations, and context-dependent phase behavior
Mean hydrophobicityHigh; supports formation of stable hydrophobic coresLow; depletion of bulky hydrophobic residuesReduces hydrophobic collapse and stable folding while permitting exposure of cryptic aggregation-prone segments
β-sheet propensityModerate to high in structured β-domainsGenerally low overall but contains localized β-prone motifsEmbedded β-prone regions can nucleate aggregation following conformational perturbation
Sequence complexityHigh sequence complexity with diverse amino acid compositionLow-complexity and compositionally biased sequencesFacilitates multivalent weak interactions associated with phase separation and dynamic assembly formation
Predominant residue typesHydrophobic and structure-promoting residuesPolar, charged, and glycine-rich residues (e.g., Glu, Lys, Arg, Gln, Ser, Gly)Promotes conformational flexibility and transient intermolecular interactions
Charge patterningLess influential due to stable tertiary structureCritical determinant of chain compaction and intermolecular interactionsRegulates conformational ensemble behavior, condensate formation, and aggregation propensity
Structural organizationStable tertiary structure with defined folding architectureDynamic conformational ensembles lacking stable tertiary structureEnables rapid conformational sampling and adaptive molecular recognition
Aggregation-prone regionsTypically buried within hydrophobic coresFrequently cryptic and embedded within disordered sequencesExposure of APRs under stress or environmental perturbation promotes aggregation without global unfolding

In contrast to folded proteins that occupy well-defined energy minima, IDRs sample heterogeneous ensembles distributed across shallow and highly dynamic energy landscapes (Lotthammer et al., 2024; Strodel, 2021; Novak et al., 2026). These ensembles continuously interconvert between expanded coil-like states and partially compact conformations stabilized by transient electrostatic interactions, hydrogen bonding, cation–π interactions, and aromatic stacking (Strodel, 2021; Novak et al., 2026; Ghafouri et al., 2026). Experimental studies using nuclear magnetic resonance (NMR), single-molecule spectroscopy, and MD simulations have shown that disordered proteins frequently retain residual secondary structural elements, including transient α-helices and β-strands, which can serve as nucleation sites for intermolecular assembly and aggregation (Novak et al., 2026; Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026). The populations of these transient conformers are strongly influenced by ionic strength, pH, macromolecular crowding, and post-translational modifications (Ahmad et al., 2022; Novak et al., 2026; Hollingsworth and Dror, 2018).

Sequence patterning within IDRs plays a critical role in defining ensemble behavior and intermolecular interaction propensity (Bremer et al., 2022; Ginell et al., 2025). Even when overall amino acid composition remains similar, variations in the spatial distribution of charged, aromatic, and hydrophobic residues can substantially alter chain compaction, phase separation behavior, and aggregation susceptibility (Bremer et al., 2022). Uniformly distributed charges generally favor expanded conformations through electrostatic repulsion, whereas block-like arrangements of oppositely charged residues facilitate transient intramolecular contacts and partial collapse (Uversky, 2021; Bremer et al., 2022). Similarly, clusters of aromatic residues enhance multivalent intermolecular interactions associated with LLPS and condensate formation (Visser et al., 2024; Banani et al., 2017; Bremer et al., 2022). These observations support the concept of “sequence grammar,” in which residue arrangement, rather than composition alone, determines the physicochemical behavior of disordered proteins (Bremer et al., 2022; Ginell et al., 2025).

An important characteristic of IDRs is the presence of cryptic APRs embedded within low-complexity sequences (Santos et al., 2021). Under physiological conditions, these APRs are frequently masked by surrounding charged and polar residues that maintain solubility and limit intermolecular association (Santos et al., 2021; Strodel, 2021). However, environmental perturbations, mutations, altered translation kinetics, or post-translational modifications can shift ensemble populations toward aggregation-competent conformations (Santos et al., 2021; Ahmad et al., 2022; Uversky, 2007). Such transitions are particularly relevant in proteins associated with neurodegenerative diseases, including α-synuclein, tau, FUS, and TDP-43, where disorder-to-order transitions promote oligomerization and fibril formation (Ahmad et al., 2022; Uversky, 2007; Utami et al., 2025; Ishiguro et al., 2021; Hayes and Kalab, 2022).

Modern computational frameworks have improved the characterization of IDR conformational landscapes through integration of MD simulations, ensemble-based predictors, and protein language model embeddings (Lotthammer et al., 2024; Novak et al., 2026; Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026). Ensemble-aware approaches now provide more accurate representations of dynamic disorder and transient intermolecular interactions than traditional static structural models (Lotthammer et al., 2024; Ghosh et al., 2024; Ghafouri et al., 2026). Collectively, these findings establish that intrinsic disorder represents a highly organized and evolutionarily tuned conformational state in which sequence composition, residue patterning, and environmental context cooperatively regulate solubility, molecular recognition, phase behavior, and aggregation propensity.

3 Molecular basis of protein aggregation in disordered contexts

Protein aggregation in IDRs is governed by the interplay between sequence-encoded APRs, conformational ensemble dynamics, and environmental conditions that regulate intermolecular interactions (Santos et al., 2021; Strodel, 2021; Ahmad et al., 2022). Unlike folded proteins, where aggregation generally follows partial unfolding and exposure of buried hydrophobic cores, aggregation in IDRs emerges from transient ensemble states that expose hydrophobic or β-sheet-prone motifs (Santos et al., 2021; Strodel, 2021). These minor-population conformers can nucleate intermolecular association, initiating the transition from soluble monomers to oligomeric intermediates and fibrillar assemblies (Strodel, 2021; Uversky, 2007).

Aggregation in disordered systems commonly follows a nucleation–elongation mechanism in which the formation of an energetically unfavorable nucleus constitutes the rate-limiting step (Strodel, 2021; Ahmad et al., 2022). In IDRs, the frequency with which aggregation-competent conformations are sampled strongly influences nucleation kinetics (Lotthammer et al., 2024; Strodel, 2021). Nucleation typically involves the transient association of partially structured monomers into small oligomeric assemblies that overcome the energetic barrier separating soluble and aggregated states. These early oligomers may subsequently reorganize into more stable β-sheet-rich nuclei capable of supporting rapid fibril growth through monomer addition during the elongation phase. The duration of the nucleation phase therefore strongly influences aggregate formation rates, growth kinetics, and fibrillation behavior, whereas elongation proceeds through cooperative intermolecular interactions that stabilize ordered aggregate architectures (Jarrett et al., 1993; Ahanger et al., 2021; Kane et al., 2000). Conformational flexibility can therefore exert dual effects: increased sampling enhances the probability of productive intermolecular encounters, whereas electrostatic repulsion and solubilizing polar interactions suppress stable assembly formation (Uversky, 2021; Strodel, 2021). These oligomeric intermediates are frequently heterogeneous and dynamic, ranging from loosely associated assemblies to partially ordered β-rich structures (Ahmad et al., 2022; Chakraborty et al., 2024). Several studies indicate that such oligomeric species exhibit greater cellular toxicity than mature fibrils due to membrane disruption, sequestration of functional proteins, and perturbation of proteostasis networks (Uversky, 2007; Chakraborty et al., 2024). In addition to primary nucleation, aggregation can be accelerated through seeded polymerization mechanisms in which pre-existing oligomeric or fibrillar fragments act as templates for monomer recruitment, effectively bypassing the nucleation barrier and promoting rapid aggregate propagation. Such seeding processes have been extensively documented for α-synuclein and are believed to contribute to the progressive spread of pathological protein assemblies in neurodegenerative disorders (Shahnawaz et al., 2020; Ahanger et al., 2023).

The aggregation propensity of IDRs is highly context dependent because APR accessibility is regulated by sequence patterning and intramolecular interactions (Santos et al., 2021; Bremer et al., 2022; Ginell et al., 2025). Short hydrophobic or β-sheet-forming motifs are often flanked by charged residues that maintain solubility under physiological conditions (Santos et al., 2021). Perturbations such as mutations, phosphorylation, acetylation, altered pH, oxidative stress, or elevated local concentration can destabilize these protective interactions and expose APRs (Ahmad et al., 2022; Utami et al., 2025; Ishiguro et al., 2021). This process is evident in disease-associated proteins including α-synuclein, tau, FUS, and TDP-43, where conformational transitions stabilize intermolecular β-sheet interactions and promote fibrillization (Uversky, 2007; Utami et al., 2025; Ishiguro et al., 2021; Hayes and Kalab, 2022; Chakraborty et al., 2024; Chen et al., 2024). Phosphorylation-dependent structural rearrangements in tau and dysregulated phase transitions in FUS provide representative examples of how post-translational modifications influence aggregation pathways (Ishiguro et al., 2021; Chen et al., 2024).

A characteristic feature of aggregation in IDRs is the disorder-to-order transition accompanying structural maturation (Strodel, 2021; Ahmad et al., 2022). Transient secondary structural elements present within conformational ensembles become stabilized upon intermolecular association, enabling cooperative assembly into cross-β architectures (Strodel, 2021; Uversky, 2007). Environmental conditions including ionic strength, molecular crowding, and temperature modulate these transitions by altering electrostatic screening and chain compaction (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022). In cellular environments, molecular chaperones and proteostasis systems further regulate aggregation kinetics by shielding APRs, promoting disassembly, or targeting aberrant assemblies for degradation (Alberti et al., 2025).

Several computational studies integrating MD simulations, ensemble-based modeling, and machine learning approaches have improved characterization of aggregation pathways in disordered systems (Novak et al., 2026; Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026; Robustelli et al., 2018). Ensemble-aware frameworks provide mechanistic insights into transient aggregation-competent states that are not captured by static structural analyses (Lotthammer et al., 2024; Novak et al., 2026; Ghosh et al., 2024). These developments support the view that aggregation in IDRs is an emergent property arising from the coupling between sequence grammar, conformational heterogeneity, and environmental regulation.

4 Interplay between intrinsic disorder, phase separation, and aggregation

IDRs play a central role in LLPS through their ability to mediate weak, multivalent intermolecular interactions (Uversky, 2021; Vendruscolo and Fuxreiter, 2022; Banani et al., 2017; Bremer et al., 2022; Alberti et al., 2025; Alberti and Hyman, 2021). Unlike structured domains that rely on highly specific binding interfaces, IDRs promote condensate formation through distributed interaction motifs involving electrostatic interactions, π–π stacking, cation–π contacts, and transient hydrophobic associations (Vendruscolo and Fuxreiter, 2022; Banani et al., 2017; Bremer et al., 2022). Low-complexity domains enriched in glycine, serine, glutamine, tyrosine, and arginine are particularly effective in driving phase separation because their conformational flexibility enables rapid assembly and disassembly of dynamic interaction networks (Vendruscolo and Fuxreiter, 2022; Banani et al., 2017). The sequence grammar of IDRs, including charge distribution, aromatic residue spacing, and hydrophobic patterning, therefore determines the balance between solubility, condensation, and aggregation propensity (Bremer et al., 2022; Ginell et al., 2025).

LLPS and aggregation are increasingly recognized as interconnected states within a broader continuum of biomolecular self-organization (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Banani et al., 2017; Alberti et al., 2025). Liquid-like condensates are generally characterized by reversible molecular exchange and structural heterogeneity, whereas aggregated assemblies exhibit increased rigidity, β-sheet enrichment, and kinetic stabilization (Ahmad et al., 2022; Alberti et al., 2025). Elevated local concentrations within condensates substantially increase the probability of intermolecular encounters between APRs, thereby lowering nucleation barriers for fibril formation (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022). Under prolonged residence times or stress conditions, initially dynamic condensates can progressively mature into gel-like or amyloid-like assemblies through disorder-to-order transitions and stabilization of intermolecular β-sheet interactions (Ahmad et al., 2022; Alberti et al., 2025; Chen et al., 2024). This maturation process has been observed in several disease-associated proteins, including α-synuclein, tau, FUS, and TDP-43 (Uversky, 2007; Utami et al., 2025; Ishiguro et al., 2021; Hayes and Kalab, 2022; Chakraborty et al., 2024; Shahnawaz et al., 2020; Ahanger et al., 2023; Chen et al., 2024).

The relationship between LLPS and pathological aggregation is strongly influenced by environmental and cellular factors (Ahmad et al., 2022; Alberti et al., 2025; Alberti and Hyman, 2021). Changes in pH, ionic strength, temperature, oxidative stress, and macromolecular crowding alter electrostatic screening and interaction lifetimes within condensates, thereby modulating their material properties and aggregation susceptibility (Vendruscolo and Fuxreiter, 2022; Banani et al., 2017). Post-translational modifications (PTMs), including phosphorylation, acetylation, and methylation, further regulate condensate dynamics by modifying charge distribution and intermolecular interaction networks (Ishiguro et al., 2021; Chakraborty et al., 2024). Phosphorylation-dependent acceleration of tau aggregation and mutation-induced dysregulation of FUS phase transitions illustrate how altered sequence or PTM states can shift condensates toward irreversible aggregation pathways (Ishiguro et al., 2021; Chakraborty et al., 2024).

Cellular proteostasis mechanisms provide additional regulation of the phase separation–aggregation balance. Molecular chaperones, degradation pathways, and RNA-binding partners can stabilize reversible condensates, shield APRs, or promote disassembly of aberrant assemblies (Alberti et al., 2025; Hayes and Kalab, 2022). Failure of these regulatory systems contributes to persistent condensate maturation and accumulation of pathogenic aggregates in neurodegenerative disorders (Uversky, 2007; Utami et al., 2025; Hayes and Kalab, 2022). Importantly, the distinction between functional condensates and pathological aggregates is not absolute but reflects quantitative differences in interaction strength, valency, molecular mobility, and reversibility (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Banani et al., 2017; Alberti et al., 2025).

Current state-of-the-art computational and experimental developments have improved mechanistic understanding of this continuum through integration of molecular simulations, cryo-electron microscopy, single-molecule methods, and ensemble-based predictive frameworks (Lotthammer et al., 2024; Novak et al., 2026; Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026; Robustelli et al., 2018). These studies demonstrate that LLPS and aggregation are governed by overlapping sequence determinants and conformational landscapes rather than independent molecular processes (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Bremer et al., 2022). Consequently, understanding disorder-driven aggregation requires a unified framework that links sequence grammar, conformational ensembles, condensate dynamics, and fibrillar maturation across multiple spatial and temporal scales.

5 Computational prediction of aggregation in intrinsically disordered proteins

Prediction of aggregation in IDPs remains challenging because aggregation propensity emerges from dynamic conformational ensembles rather than stable structural states (Uversky, 2021; Strodel, 2021; Zambrano et al., 2015; Ginell et al., 2025; Novak et al., 2026; Dokainish et al., 2022). Early computational approaches were primarily developed for folded proteins and therefore relied heavily on hydrophobicity, β-sheet propensity, and residue packing energies (Fernandez-Escamilla et al., 2004; Walsh et al., 2014; Tartaglia and Vendruscolo, 2008). Widely used tools such as TANGO, PASTA, Zyggregator, and AGGRESCAN estimate aggregation potential based on physicochemical properties associated with amyloid formation and intermolecular β-sheet stabilization (Fernandez-Escamilla et al., 2004; Walsh et al., 2014; Tartaglia and Vendruscolo, 2008; Zambrano et al., 2015). TANGO predicts β-aggregation nucleation using statistical thermodynamic principles, whereas PASTA evaluates inter-strand pairing energies and β-sheet compatibility (Fernandez-Escamilla et al., 2004; Walsh et al., 2014). Zyggregator integrates hydrophobicity, charge, and secondary structure propensity to estimate aggregation likelihood, while AGGRESCAN and AGGRESCAN3D identify APRs from sequence and structural contexts (Tartaglia and Vendruscolo, 2008; Zambrano et al., 2015). These approaches typically generate residue-level or sequence-level aggregation propensity scores that can be used to identify aggregation hotspots, estimate amyloid-forming potential, and prioritize regions for experimental characterization. Although these methods remain effective for globular proteins, their performance is often reduced for IDRs because they inadequately capture ensemble heterogeneity, transient structural states, and context-dependent APR masking (Santos et al., 2021; Strodel, 2021). Classical aggregation predictors, disorder-aware methods, and emerging ensemble- and protein language model-based frameworks relevant to IDR aggregation analysis are compared in Table 2.

TABLE 2. Computational predictors and emerging frameworks for aggregation analysis in intrinsically disordered proteins. Computational approaches for predicting aggregation propensity range from classical physicochemical models to modern machine learning and protein language model (PLM)-based frameworks. More recent disorder-aware approaches incorporate conformational ensemble behavior, charge distribution, sequence complexity, and contextual sequence embeddings to better characterize aggregation in intrinsically disordered regions (IDRs).

Tool/FrameworkInput data typeCore prediction principleDisorder-aware featuresApplication and output
TANGOProtein sequence with environmental parameters (pH, temperature, ionic strength)Statistical thermodynamic model of β-sheet aggregation and competing secondary structure formationLimited; indirectly considers environmental effects influencing conformational statesPrediction of β-aggregation nucleation and amyloid propensity through residue-level and sequence-level aggregation scores
AGGRESCAN/AGGRESCAN3DProtein sequence or structural modelEmpirical aggregation propensity scales derived from experimental aggregation dataPartial; identifies aggregation-prone regions but limited explicit disorder modelingDetection of aggregation-prone regions (APRs), aggregation hotspots, and local aggregation propensity profiles
ZyggregatorProtein sequencePhysicochemical model integrating hydrophobicity, charge, and β-sheet propensityPartial; incorporates charge effects but limited ensemble representationGlobal aggregation propensity estimation and ranking of aggregation susceptibility
WaltzProtein sequencePosition-specific scoring matrix optimized for amyloid-forming motifsLimited; motif-focused without broader disorder contextIdentification of short amyloidogenic peptide segments and sequence motifs associated with fibril formation
IUPred/PONDR/DISOPREDProtein sequenceEnergy estimation and machine learning-based disorder predictionHigh for disorder identification; indirect aggregation relevancePrediction of intrinsically disordered regions, flexible segments, and potential locations of cryptic APR exposure
Classical machine learning predictors (SVM/RF/NN-based)Protein sequence with engineered physicochemical descriptorsSupervised learning trained on experimentally annotated aggregation datasetsModerate; incorporate disorder propensity, charge distribution, and sequence complexityAggregation classification, APR prediction, and quantitative aggregation propensity estimation
Ensemble-based molecular dynamics frameworksProtein sequence and structural ensemblesSimulation of conformational landscapes and transient intermolecular interactionsHigh; captures ensemble heterogeneity and disorder-to-order transitionsAnalysis of aggregation pathways, condensate maturation, conformational dynamics, and transient aggregation-competent states
Protein language model (PLM)-based methodsProtein sequence embeddings from large-scale protein databasesDeep learning models capturing contextual sequence relationships and long-range dependenciesHigh; captures latent sequence features linked to disorder and aggregationContext-aware aggregation prediction, identification of cryptic aggregation-prone motifs, interaction propensity estimation, and sequence-function modeling

To address these limitations, newer approaches incorporate disorder-specific sequence features including net charge per residue, sequence complexity, aromatic residue distribution, and charge patterning (Uversky, 2021; Bremer et al., 2022; Zambrano et al., 2015). Predictors such as IUPred, PONDR, DISOPRED, and flDPnn are frequently used to identify intrinsically disordered segments prior to aggregation analysis (Uversky, 2021; Lotthammer et al., 2024). Sequence grammar analysis has further improved understanding of how residue spacing and multivalent interaction motifs regulate phase behavior and aggregation susceptibility (Bremer et al., 2022). Several studies have demonstrated that APR accessibility in IDRs depends not only on amino acid composition but also on local sequence organization and environmental context (Santos et al., 2021; Zambrano et al., 2015).

Evolving computational approaches increasingly integrate MD simulations with machine learning and PLM-based approaches (Ginell et al., 2025; Novak et al., 2026; Kaushik and Launey, 2024; Mall et al., 2025; Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026; Kaushik and Zhang, 2020; Kaushik and Zhang, 2022a; Kaushik and Zhang, 2022b). Ensemble-based frameworks such as STARLING and related conformational prediction models estimate the dynamic structural landscapes of IDRs rather than predicting single static conformations (Ginell et al., 2025; Novak et al., 2026). These methods generate conformational ensemble descriptors that provide insight into transient APR exposure, intermolecular interaction propensity, and aggregation-competent states. The MD simulations using optimized force fields have further improved characterization of transient intermolecular interactions, condensate maturation, and disorder-to-order transitions associated with aggregation (Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026). These approaches are particularly useful for studying proteins such as α-synuclein, tau, and FUS, where aggregation is strongly coupled to conformational heterogeneity and LLPS (Ahmad et al., 2022; Uversky, 2007; Utami et al., 2025; Fernandez-Escamilla et al., 2004; Chakraborty et al., 2024; Chen et al., 2024).

Machine learning approaches now incorporate large-scale sequence embeddings, evolutionary information, and structural descriptors to improve prediction accuracy (Zambrano et al., 2015; Kaushik and Launey, 2024; Mall et al., 2025; Ahanger et al., 2023; Robustelli et al., 2018). Protein language models trained on large proteomic datasets capture long-range sequence dependencies and implicit biophysical relationships associated with disorder and aggregation (Mall et al., 2025). Rather than relying solely on predefined physicochemical descriptors, PLM-derived embeddings encode contextual sequence information that can be used to predict aggregation propensity, identify cryptic aggregation-prone motifs, estimate intermolecular interaction tendencies, and characterize sequence features associated with phase separation behavior. Benchmarking studies indicate that PLM-derived embeddings can outperform conventional handcrafted feature approaches in identifying aggregation-prone motifs and conformational tendencies (Mall et al., 2025). Hybrid frameworks combining sequence analysis, structural fitness scoring, and ensemble-aware modeling have also shown improved predictive performance for complex disordered systems (Kaushik and Launey, 2024; Kaushik and Zhang, 2020; Kaushik and Zhang, 2022a; Kaushik and Zhang, 2022b).

6 Biological implications

Aggregation of IDPs has significant implications for cellular homeostasis, neurodegeneration, and protein quality control systems (Ahmad et al., 2022; Uversky, 2007; Utami et al., 2025; Hayes and Kalab, 2022; Chakraborty et al., 2024; Shahnawaz et al., 2020; Ahanger et al., 2023; Chen et al., 2024). Many disease-associated proteins, including α-synuclein, tau, FUS, and TDP-43, contain extensive IDRs that facilitate functional interactions under physiological conditions but also increase susceptibility to aberrant self-assembly (Uversky, 2007; Utami et al., 2025; Hayes and Kalab, 2022; Hayes and Kalab, 2022; Chakraborty et al., 2024; Chen et al., 2024). Representative disease-associated intrinsically disordered proteins that undergo transitions from functional condensates to pathological aggregates are summarized in Table 3. The transition from dynamic conformational ensembles to oligomeric and fibrillar aggregates disrupts multiple cellular processes, including RNA metabolism, membrane integrity, intracellular transport, and stress response pathways (Utami et al., 2025; Hayes and Kalab, 2022; Jarrett et al., 1993). Accumulation of such assemblies is a defining pathological feature of several neurodegenerative disorders, including Parkinson’s disease, Alzheimer’s disease, amyotrophic lateral sclerosis (ALS), and frontotemporal dementia (FTD) (Uversky, 2007; Utami et al., 2025; Hayes and Kalab, 2022; Chakraborty et al., 2024; Chen et al., 2024).

TABLE 3. Phase separation–aggregation continuum of disease-associated intrinsically disordered proteins. Several intrinsically disordered proteins (IDPs) associated with neurodegenerative diseases undergo transitions from functional phase-separated condensates to irreversible pathological aggregates. Intrinsically disordered regions facilitate multivalent intermolecular interactions that support liquid–liquid phase separation (LLPS) and formation of dynamic biomolecular condensates under physiological conditions.

Protein/Disease associationFunctional condensate or physiological stateCondensate maturation/Aging mechanismPathological aggregateMajor sequence or structural determinants
α-Synuclein/Parkinson’s diseaseSynaptic vesicle-associated assemblies; transient membrane-bound condensatesConcentration-dependent condensation, lipid-mediated structural rearrangement, and condensate compactionAmyloid fibrils and Lewy bodiesHydrophobic NAC region, KTKEGV repeats, charge distribution influencing membrane interaction
Tau/Alzheimer’s diseaseMicrotubule-associated condensates and RNA-mediated dropletsPhosphorylation-induced charge redistribution, reduced microtubule binding, and enhanced intermolecular associationPaired helical filaments and neurofibrillary tanglesRepeat domains (R2–R4), PHF6/PHF6* motifs, phosphorylation sites
FUS/Amyotrophic lateral sclerosis (ALS)Nuclear RNA–protein granules and stress granulesLiquid-to-gel transition, mutation-dependent stabilization of intermolecular interactions, and altered nucleocytoplasmic localizationCytoplasmic fibrillar aggregatesLow-complexity prion-like domain, tyrosine-mediated π interactions, arginine-rich regions
TDP-43/ALS and frontotemporal lobar degeneration (FTLD)Stress granules and RNA-binding condensatesC-terminal domain-mediated phase separation, post-translational modifications, and proteolytic processingCytoplasmic inclusions and amyloid-like aggregatesGlycine-rich C-terminal domain, glutamine/asparagine-rich segments, phosphorylation and ubiquitination sites

Increasing evidence indicates that oligomeric intermediates formed during aggregation frequently exhibit greater cytotoxicity than mature amyloid fibrils (Strodel, 2021; Ahmad et al., 2022). These metastable assemblies interact with cellular membranes, alter ion homeostasis, and interfere with proteostasis networks (Strodel, 2021; Uversky, 2007). Dysregulated LLPS further contributes to disease progression by promoting condensate maturation and persistence of aggregation-prone states (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Banani et al., 2017; Alberti et al., 2025). Cellular stress conditions, impaired chaperone activity, oxidative damage, and altered post-translational modifications can shift the balance from reversible condensates toward irreversible fibrillar assemblies (Ahmad et al., 2022; Alberti et al., 2025; Alberti and Hyman, 2021). Consequently, maintenance of condensate dynamics and efficient protein quality control are critical for preserving cellular function.

Beyond disease biology, intrinsic disorder and regulated self-assembly also contribute to normal cellular organization and adaptive responses (Csizmok et al., 2016; Uversky, 2021; Banani et al., 2017). IDRs mediate transient interactions involved in signaling, transcriptional regulation, stress granule formation, and biomolecular compartmentalization (Csizmok et al., 2016; Vendruscolo and Fuxreiter, 2022; Banani et al., 2017). The same physicochemical principles governing pathological aggregation therefore also involved in functional molecular assembly processes. Understanding how sequence grammar, environmental conditions, and conformational landscapes distinguish functional condensates from pathogenic aggregates remains a major challenge in molecular biophysics (Vendruscolo and Fuxreiter, 2022; Bremer et al., 2022).

7 Emerging directions in disorder-aware aggregation research

Modern computational frameworks together with advances in experimental biophysics are substantially improving mechanistic understanding of aggregation in IDPs (Ginell et al., 2025; Novak et al., 2026; Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026). Although considerable progress has been made in identifying aggregation-prone regions and characterizing conformational ensembles, accurately predicting aggregation behavior under physiologically relevant conditions remains a major challenge. Emerging disorder-aware frameworks are increasingly moving beyond static sequence-based predictions toward integrative models that incorporate conformational dynamics, sequence grammar, phase behavior, and cellular context (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Banani et al., 2017; Bremer et al., 2022; Alberti et al., 2025).

One promising direction is the development of multimodal artificial intelligence frameworks capable of integrating sequence information, conformational ensemble descriptors, structural data, and experimental measurements within unified predictive architectures (Ginell et al., 2025; Novak et al., 2026; Kaushik and Launey, 2024; Mall et al., 2025). Such approaches may improve identification of aggregation-competent states that arise only under specific environmental or cellular conditions. In parallel, growing availability of large-scale proteomic, structural, and biophysical datasets is expected to facilitate development of more interpretable and biologically informed predictive models (Ginell et al., 2025; Mall et al., 2025; Kumar et al., 2021b).

Parallel developments in MD simulations, cryo-electron microscopy, and integrative structural biology are enabling improved characterization of condensate maturation and disorder-to-order transitions (Ghosh et al., 2024; Hollingsworth and Dror, 2018; Ghafouri et al., 2026). Enhanced force fields optimized for both folded and disordered proteins now allow more accurate simulation of conformational ensembles and intermolecular assembly pathways (Ghosh et al., 2024; Hollingsworth and Dror, 2018). Integration of experimental methods with multiscale computational simulations is therefore becoming increasingly important for studying the coupling between LLPS, oligomerization, and fibrillization (Ahmad et al., 2022; Vendruscolo and Fuxreiter, 2022; Alberti et al., 2025).

Despite these advances, important limitations remain. Most predictive models still inadequately represent environmental variables such as molecular crowding, RNA interactions, post-translational modifications, and condensate-specific physicochemical conditions (Ahmad et al., 2022; Alberti et al., 2025; Alberti and Hyman, 2021; Kasahara et al., 2021). Furthermore, experimentally validated datasets for IDR aggregation remain limited compared with datasets for folded proteins (Zambrano et al., 2015; Ginell et al., 2025). Future progress will likely depend on integration of experimental biophysics, cryo-electron microscopy, MD simulations, and interpretable artificial intelligence frameworks capable of modeling dynamic conformational ensembles under physiologically relevant conditions (Ginell et al., 2025; Novak et al., 2026; Ghosh et al., 2024; Ghafouri et al., 2026; Kumar et al., 2023; Kaushik et al., 2024; Kaushik et al., 2022). Applications of artificial intelligence in computational protein engineering and vaccine design further demonstrate the growing utility of data-driven frameworks for predictive biomolecular modeling and therapeutic development (Kaushik and Re, 2026; Kaushik et al., 2023). Such integrative approaches are expected to improve aggregation prediction, facilitate identification of disease-associated mutations, support rational design of aggregation-resistant proteins, and ultimately accelerate development of targeted therapeutic strategies.

8 Conclusion

IDRs represent dynamic structural elements that play fundamental roles in molecular recognition, signaling, and biomolecular assembly while simultaneously contributing to protein aggregation susceptibility. Unlike aggregation in folded proteins, aggregation in IDPs emerges from transient conformational ensembles in which cryptic APRs become exposed under specific environmental and cellular conditions. Increasing evidence demonstrates that intrinsic disorder, LLPS, and amyloid formation are mechanistically interconnected processes governed by sequence grammar, intermolecular interactions, and conformational heterogeneity.

The progression from reversible condensate formation to irreversible fibrillar aggregation is particularly relevant in neurodegenerative disorders involving α-synuclein, tau, FUS, and TDP-43. These transitions are strongly influenced by post-translational modifications, molecular crowding, cellular stress, and impaired proteostasis mechanisms. Consequently, understanding aggregation in disordered systems requires integrated analysis of sequence composition, ensemble dynamics, and cellular context.

Current developments in MD simulations, ensemble-aware modeling, and PLMs-based approaches have improved the prediction and characterization of disorder-driven aggregation pathways. However, accurate representation of dynamic intracellular environments and transient assembly states remains a major challenge. Continued integration of experimental biophysics with multiscale computational frameworks will be essential for improving aggregation prediction, understanding disease-associated self-assembly, and developing targeted therapeutic strategies.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Chiranjeevi Pasala, Sloan Kettering Insitute, United States

Reviewed by: Ishfaq A. Ahanger, University of Kashmir, India

Samuel Wohl, Arizona State University - Polytechnic Campus, United States

Statements

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author RK declared that they were an editorial board member of Frontiers at the time of submission. This had no impact on the peer review process and the final decision.

Funding

The author(s) declared that financial support was not received for this work and/or its publication.

Author contributions

RK: Investigation, Conceptualization, Writing – review and editing, Formal Analysis, Writing – original draft. SR: Writing – review and editing, Investigation, Funding acquisition.

Acknowledgements

The figure in this review article was created with BioRender.com.

References

  1. Ahanger I. A., Bashir S., Parray Z. A., Alajmi M. F., Hussain A., Ahmad F. (2021). Rationalizing the role of monosodium glutamate in the protein aggregation through biophysical approaches: potential impact on neurodegeneration. Front. Neurosci. 15, 636454. doi:10.3389/fnins.2021.636454
  2. Ahanger I. A., Parray Z. A., Raina N., Bashir S., Ahmad F., Hassan M. I. (2023). Counteraction of the cetyltrimethylammonium bromide-induced protein aggregation by heparin: potential impact on protein aggregation and neurodegenerative diseases using biophysical approaches. J. Mol. Struct. 1278, 134714. doi:10.1016/j.molstruc.2022.134714
  3. Ahmad A., Uversky V. N., Khan R. H. (2022). Aberrant liquid-liquid phase separation and amyloid aggregation of proteins related to neurodegenerative diseases. Int. J. Biol. Macromol. 220, 703 – 720. doi:10.1016/j.ijbiomac.2022.08.132
  4. Alberti S., Hyman A. A. (2021). Biomolecular condensates at the nexus of cellular stress, protein aggregation disease and ageing. Nat. Rev. Mol. Cell Biol. 22 (3), 196 – 213. doi:10.1038/s41580-020-00326-6
  5. Alberti S., Arosio P., Best R. B., Boeynaems S., Cai D., Collepardo-Guevara R. (2025). Current practices in the study of biomolecular condensates: a community comment. Nat. Commun. 16 (1), 7730. doi:10.1038/s41467-025-62055-8
  6. Banani S. F., Lee H. O., Hyman A. A., Rosen M. K. (2017). Biomolecular condensates: organizers of cellular biochemistry. Nat. Rev. Mol. Cell Biol. 18 (5), 285 – 298. doi:10.1038/nrm.2017.7
  7. Bremer A., Farag M., Borcherds W. M., Peran I., Martin E. W., Pappu R. V. (2022). Deciphering how naturally occurring sequence features impact the phase behaviours of disordered prion-like domains. Nat. Chem. 14 (2), 196 – 207. doi:10.1038/s41557-021-00840-w
  8. Chakraborty P., Ibáñez De Opakua A., Purslow J. A., Fromm S. A., Chatterjee D., Zachrdla M. (2024). GSK3β phosphorylation catalyzes the aggregation of tau into Alzheimer’s disease-like filaments. Proc. Natl. Acad. Sci. U. S. A. 121 (52), e2414176121. doi:10.1073/pnas.2414176121
  9. Chen S. W., Barritt J. D., Cascella R., Bigi A., Cecchi C., Banchelli M. (2024). Structure–toxicity relationship in intermediate fibrils from α-Synuclein condensates. J. Am. Chem. Soc. 146 (15), 10537 – 10549. doi:10.1021/jacs.3c14703
  10. Csizmok V., Follis A. V., Kriwacki R. W., Forman-Kay J. D. (2016). Dynamic protein interaction networks and new structural paradigms in signaling. Chem. Rev. 116 (11), 6424 – 6462. doi:10.1021/acs.chemrev.5b00548
  11. Dokainish H. M., Re S., Mori T., Kobayashi C., Jung J., Sugita Y. (2022). The inherent flexibility of receptor binding domains in SARS-CoV-2 spike protein. eLife 11, e75720. doi:10.7554/eLife.75720
  12. Fernandez-Escamilla A.-M., Rousseau F., Schymkowitz J., Serrano L. (2004). Prediction of sequence-dependent and mutational effects on the aggregation of peptides and proteins. Nat. Biotechnol. 22 (10), 1302 – 1306. doi:10.1038/nbt1012
  13. Ghafouri H., Tosatto S. C. E., Monzon A. M. (2026). Advances in the determination of disordered protein ensemble. Curr. Opin. Struct. Biol. 96, 103198. doi:10.1016/j.sbi.2025.103198
  14. Ghosh C., Nagpal S., Muñoz V. (2024). Molecular simulations integrated with experiments for probing the interaction dynamics and binding mechanisms of intrinsically disordered proteins. Curr. Opin. Struct. Biol. 84, 102756. doi:10.1016/j.sbi.2023.102756
  15. Ginell G. M., Emenecker R. J., Lotthammer J. M., Keeley A. T., Plassmeyer S. P., Razo N. (2025). Sequence-based prediction of intermolecular interactions driven by disordered regions. Science 388 (6749), eadq8381. doi:10.1126/science.adq8381
  16. Hayes L. R., Kalab P. (2022). Emerging therapies and novel targets for TDP-43 proteinopathy in ALS/FTD. Neurotherapeutics 19 (4), 1061 – 1084. doi:10.1007/s13311-022-01260-5
  17. Hollingsworth S. A., Dror R. O. (2018). Molecular dynamics simulation for all. Neuron 99 (6), 1129 – 1143. doi:10.1016/j.neuron.2018.08.011
  18. Ishiguro A., Lu J., Ozawa D., Nagai Y., Ishihama A. (2021). ALS-Linked FUS mutations dysregulate G-quadruplex-dependent liquid–liquid phase separation and liquid-to-solid transition. J. Biol. Chem. 297 (5), 101284. doi:10.1016/j.jbc.2021.101284
  19. Jarrett J. T., Berger E. P., Lansbury P. T. Jr (1993). The carboxy terminus of the beta amyloid protein is critical for the seeding of amyloid formation: implications for the pathogenesis of Alzheimer's disease. Biochemistry 32 (18), 4693 – 4697. doi:10.1021/bi00069a001
  20. Kane M. D., Lipinski W. J., Callahan M. J., Bian F., Durham R. A., Schwarz R. D. (2000). Evidence for seeding of beta -amyloid by intracerebral infusion of Alzheimer brain extracts in beta-amyloid precursor protein-transgenic mice. J. Neurosci. 20 (10), 3606 – 3611. doi:10.1523/JNEUROSCI.20-10-03606.2000
  21. Kasahara K., Re S., Nawrocki G., Oshima H., Mishima-Tsumagari C., Miyata-Yabuki Y. (2021). Reduced efficacy of a Src kinase inhibitor in crowded protein solution. Nat. Commun. 12 (1), 4099. doi:10.1038/s41467-021-24349-5
  22. Kaushik R., Launey T. (2024). Decoding Protein Aggregation Through Computational Approach: Identification and Scoring of Aggregation-Prone Regions in Protein Sequences. doi:10.1101/2024.06.11.598423
  23. Kaushik R., Re S. (2026). Artificial intelligence directed computational protein design: lessons from COVID-19 for pandemic-ready vaccines and antibody therapeutics. J. Pharm. Pharm. Sci. 29, 16146. doi:10.3389/jpps.2026.16146
  24. Kaushik R., Zhang K. Y. J. (2020). A protein sequence fitness function for identifying natural and nonnatural proteins. Proteins 88 (10), 1271 – 1284. doi:10.1002/prot.25900
  25. Kaushik R., Zhang K. Y. J. (2022a). An integrated protein structure fitness scoring approach for identifying native-like model structures. Comput. Struct. Biotechnol. J. 20, 6467 – 6472. doi:10.1016/j.csbj.2022.11.032
  26. Kaushik R., Zhang K. Y. J. (2022b). ProFitFun: a protein tertiary structure fitness function for quantifying the accuracies of model structures. GBioinformatics 38 (2), 369 – 376. doi:10.1093/bioinformatics/btab666
  27. Kaushik R., Kumar N., Zhang K. Y. J., Srivastava P., Bhatia S., Malik Y. S. (2022). A novel structure-based approach for identification of vertebrate susceptibility to SARS-CoV-2: implications for future surveillance programmes. Environ. Res. 212, 113303. doi:10.1016/j.envres.2022.113303
  28. Kaushik R., Kant R., Christodoulides M. (2023). Artificial intelligence in accelerating vaccine development - current and future perspectives. Front. Bacteriol. 2, 1258159. doi:10.3389/fbrio.2023.1258159
  29. Kaushik R., Kumar N., Launey T. (2024). “ A high-throughput computational pipeline for selection of effective antibody therapeutics against viruses,” in Advances in Antiviral Research. Editors Kumar N., Malik Y. S., Tomar S., Ezzikouri S., 327 – 344. doi:10.1007/978-981-99-9195-2_13
  30. Kumar N., Kaushik R., Tennakoon C., Uversky V. N., Longhi S., Zhang K. Y. J. (2021a). Insights into the evolutionary forces that shape the codon usage in the viral genome segments encoding intrinsically disordered protein regions. Brief. Bioinform 22 (5), bbab145. doi:10.1093/bib/bbab145
  31. Kumar N., Kaushik R., Tennakoon C., Uversky V. N., Longhi S., Zhang K. Y. J. (2021b). Comprehensive intrinsic disorder analysis of 6108 viral proteomes: from the extent of intrinsic disorder penetrance to functional annotation of disordered viral proteins. J. Proteome Res. 20 (5), 2704 – 2713. doi:10.1021/acs.jproteome.1c00011
  32. Kumar N., Kaushik R., Zhang K. Y. J., Uversky V. N., Sahu U., Sood R. (2023). A novel consensus-based computational pipeline for screening of antibody therapeutics for efficacy against SARS‐CoV‐2 variants of concern including omicron variant. Proteins 91 (6), 798 – 806. doi:10.1002/prot.26467
  33. Kumar N., Kaushik R., Dayhoff G. W., Uversky V. N., Singh A., Bhatia S. (2025). Understanding functional flexibility from intrinsic disorder landscapes of fourteen bat species proteomes. Int. J. Biol. Macromol. 330, 148251. doi:10.1016/j.ijbiomac.2025.148251
  34. Lotthammer J. M., Ginell G. M., Griffith D., Emenecker R. J., Holehouse A. S. (2024). Direct prediction of intrinsically disordered protein conformational properties from sequence. Nat. Methods 21 (3), 465 – 476. doi:10.1038/s41592-023-02159-5
  35. Mall R., Kaushik R., Martinez Z. A., Thomson M. W., Castiglione F. (2025). Benchmarking protein language models for protein crystallization. Sci. Rep. 15 (1), 2381. doi:10.1038/s41598-025-86519-5
  36. Novak B., Lotthammer J. M., Emenecker R. J., Holehouse A. S. (2026). Accurate predictions of disordered protein ensembles with STARLING. Nature 652 (8108), 240 – 250. doi:10.1038/s41586-026-10141-2
  37. Robustelli P., Piana S., Shaw D. E. (2018). Developing a molecular dynamics force field for both folded and disordered protein states. Proc. Natl. Acad. Sci. U. S. A. 115 (21), E4758-E4766. doi:10.1073/pnas.1800690115
  38. Santos J., Pallarès I., Iglesias V., Ventura S. (2021). Cryptic amyloidogenic regions in intrinsically disordered proteins: function and disease association. Comput. Struct. Biotechnol. J. 19, 4192 – 4206. doi:10.1016/j.csbj.2021.07.019
  39. Shahnawaz M., Mukherjee A., Pritzkow S., Mendez N., Rabadia P., Liu X. (2020). Discriminating α-synuclein strains in parkinson's disease and multiple system atrophy. Nature 578 (7794), 273 – 277. doi:10.1038/s41586-020-1984-7
  40. Strodel B. (2021). Energy landscapes of protein aggregation and conformation switching in intrinsically disordered proteins. J. Mol. Biol. 433 (20), 167182. doi:10.1016/j.jmb.2021.167182
  41. Tartaglia G. G., Vendruscolo M. (2008). The Zyggregator method for predicting protein aggregation propensities. Chem. Soc. Rev. 37 (7), 1395 – 1401. doi:10.1039/b706784b
  42. Utami K. H., Morimoto S., Mitsukura Y., Okano H. (2025). The roles of intrinsically disordered proteins in neurodegeneration. Biochimica Biophysica Acta (BBA) - General Subj. 1869 (4), 130772. doi:10.1016/j.bbagen.2025.130772
  43. Uversky V. N. (2007). Neuropathology, biochemistry, and biophysics of α‐synuclein aggregation. J. Neurochem. 103 (1), 17 – 37. doi:10.1111/j.1471-4159.2007.04764.x
  44. Uversky V. N. (2021). Recent developments in the field of intrinsically disordered proteins: intrinsic disorder–based emergence in cellular biology in light of the physiological and pathological liquid–liquid phase transitions. Annu. Rev. Biophys. 50 (1), 135 – 156. doi:10.1146/annurev-biophys-062920-063704
  45. Vendruscolo M., Fuxreiter M. (2022). Sequence determinants of the aggregation of proteins within condensates generated by liquid-liquid phase separation. J. Mol. Biol. 434 (1), 167201. doi:10.1016/j.jmb.2021.167201
  46. Visser B. S., Lipiński W. P., Spruijt E. (2024). The role of biomolecular condensates in protein aggregation. Nat. Rev. Chem. 8 (9), 686 – 700. doi:10.1038/s41570-024-00635-w
  47. Walsh I., Seno F., Tosatto S. C. E., Trovato A. (2014). PASTA 2.0: an improved server for protein aggregation prediction. Nucleic Acids Res. 42 (W1), W301 – W307. doi:10.1093/nar/gku399
  48. Yagi K., Re S., Mori T., Sugita Y. (2022). Weight average approaches for predicting dynamical properties of biomolecules. Curr. Opin. Struct. Biol. 72, 88 – 94. doi:10.1016/j.sbi.2021.08.008
  49. Zambrano R., Jamroz M., Szczasiuk A., Pujols J., Kmiecik S., Ventura S. (2015). AGGRESCAN3D (A3D): server for prediction of aggregation properties of protein structures. Nucleic Acids Res. 43 (W1), W306 – W313. doi:10.1093/nar/gkv359

Citation

Kaushik, R. & Re, S. (2026). Intrinsically disordered regions as drivers of protein aggregation: mechanisms, phase separation, and emerging predictive frameworks. Frontiers in Biophysics, Volume 4. https://doi.org/
Received
Accepted
Published
Volume
4 - 2026

Copyright

Copyright © 2026 Kaushik and Re.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Rahul Kaushik

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

ReviewInvestigating protein conformational ensembles by native mass spectrometry and complementary biophysical techniques—the case of α-synucleinStefania Brocca, Erika Ponzini, Carlo Santambrogio, Rita GrandoriHypothesis and TheoryPiezometric architecture and the geometry of plasticity: a proposed electromechanical model of the Fibrosis–Plasticity axisAlejandro CarballoMini ReviewElectrochemical scanning tunneling microscopy as a tool for probing protein function: from electron transfer to molecular recognition and chemosensationJosep Samitier, Anna Lagunas