Dragana Dudić1*, Ana Đokić2 and Darko Grbović3
1Alfa BK University, Faculty of Mathematics and Computer Science
2Information Technology School
3Alfa BK University, Faculty of Information Technology
dragana.dudic [at] alfa.edu.rs
Abstract
Intrinsically disordered proteins (IDPs) play central roles in stress-responsive signaling and regulation in plants, yet the extent to which protein disorder is systematically associated with stress conditions across species remains unclear. Here, we investigated the relationship between intrinsic disorder and stress status in four plant species: Arabidopsis thaliana, Zea mays, Oryza sativa, and Solanum lycopersicum, with a specific focus on IDP-rich proteins.
Protein-level disorder scores were computed using IUPred, and proteins were classified as IDP-rich based on their overall disorder fraction. We compared disorder-related features between stress and non-stress conditions within each species using non-parametric tests and effect size estimates, and further assessed their predictive value using generalized linear models.
Across all four species, stress and non-stress proteins exhibited largely overlapping disorder distributions within the IDP-rich subset. While Wilcoxon tests identified statistically significant differences in some species, effect sizes were consistently small, as reflected by negligible Cliff’s delta values. Logistic regression analyses indicated that disorder fraction and proline content showed weak but statistically detectable associations with stress status, whereas protein length emerged as a stronger and more consistent predictor. Multivariate analyses did not reveal a clear separation between stress and non-stress proteins based solely on disorder-related features.
Taken together, these results suggest that, among IDP-rich proteins, intrinsic disorder alone does not robustly distinguish stress-associated proteins from non-stress proteins across plant species. Instead, protein length and compositional features appear to contribute more substantially to stress-related differences, underscoring the need to interpret disorder in the context of broader sequence properties rather than as an isolated determinant.
Keywords: disorder, proteins, stress, plants, prediction