Compositional Bias of Intrinsically Disordered Proteins and Regions and Their Predictions

Abstract

Intrinsically disordered regions (IDRs) carry out many cellular functions and vary in length and placement in protein sequences. This diversity leads to variations in the underlying compositional biases, which were demonstrated for the short vs. long IDRs. We analyze compositional biases across four classes of disorder: fully disordered proteins; short IDRs; long IDRs; and binding IDRs. We identify three distinct biases: for the fully disordered proteins, the short IDRs and the long and binding IDRs combined. We also investigate compositional bias for putative disorder produced by leading disorder predictors and find that it is similar to the bias of the native disorder. Interestingly, the accuracy of disorder predictions across different methods is correlated with the correctness of the compositional bias of their predictions highlighting the importance of the compositional bias. The predictive quality is relatively low for the disorder classes with compositional bias that is the most different from the “generic” disorder bias, while being much higher for the classes with the most similar bias. We discover that different predictors perform best across different classes of disorder. This suggests that no single predictor is universally best and motivates the development of new architectures that combine models that target specific disorder classes.

Systematic identification of 20S proteasome substrates

“…5B ). The latter are generally found at the core of a globular protein, while the former are known as hallmarks of proteins containing IDRs (Zhao and Kurgan, 2022 ). Taken together, this suggests that the lack of structure is a distinguishing feature of 20S proteasome substrates.…”

Protein structure–function continuum model: Emerging nexuses between specificity, evolution, and structure

“…This idea is further supported by the observation that the amino acid sequences of IDPs/IDRs are characterized by the presence of nonrandom binary patterns (Cohan et al, 2022). The lack of universality of the amino acid compositions of IDPs and IDRs is illustrated by the presence of noticeable differences in the amino acid compositions of long and short IDRs (Radivojac et al, 2004; Romero et al, 1997), as well as differences between fully disordered proteins, short IDRs, long IDRs, and binding IDRs (Zhao & Kurgan, 2022).…”

Computational prediction of disordered binding regions

“…Recent research has shifted from building disorder predictors to developing methods that predict binding IDRs. Similar to IDRs, recent study shows that binding IDRs also have compositional bias in their sequences [48] , suggesting that they can be predicted directly from the sequence. Significance of these predictors is reflected by the inclusion of the assessment of the binding IDRs predictions in the recently completed community-organized Critical Assessment of Intrinsic disorder (CAID) experiment [63] .…”

“…The limited collection of annotated binding IDRs can be used to develop and evaluate computational predictors, which then can be utilized to predict these regions for the millions of protein sequences that remain unannotated. This approach relies on the fact that the disordered nature of IDRs is intrinsic (i.e., encoded) in their underlying sequences 4 , 48 , 49 , 50 , making them predictable from the sequence. This has motivated development of numerous methods that accurately predict IDRs from the protein sequence 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , with over 100 methods that were developed to date [62] .…”