Challenges in describing the conformation and dynamics of proteins with ambiguous behavior

Abstract

Traditionally, our understanding of how proteins operate and how evolution shapes them is based on two main data sources: the overall protein fold and the protein amino acid sequence. However, a significant part of the proteome shows highly dynamic and/or structurally ambiguous behavior, which cannot be correctly represented by the traditional fixed set of static coordinates. Representing such protein behaviors remains challenging and necessarily involves a complex interpretation of conformational states, including probabilistic descriptions. Relating protein dynamics and multiple conformations to their function as well as their physiological context (e.g., post-translational modifications and subcellular localization), therefore, remains elusive for much of the proteome, with studies to investigate the effect of protein dynamics relying heavily on computational models. We here investigate the possibility of delineating three classes of protein conformational behavior: order, disorder, and ambiguity. These definitions are explored based on three different datasets, using interpretable machine learning from a set of features, from AlphaFold2 to sequence-based predictions, to understand the overlap and differences between these datasets. This forms the basis for a discussion on the current limitations in describing the behavior of dynamic and ambiguous proteins.

Protein structuromics: A new method for protein structure–function crosstalk in glioma

“…We followed the Grau l and Joel Roca‐Martinez publications 32,33 . For each dataset and each prediction, every amino acid was further labeled to show whether it was a single (S) early folding or nonearly folding residue or whether it appeared at the beginning (B), middle (M) or end (E) of a given early folding or nonearly folding fragment (Figure S1).…”

“…We followed the Grau l and Joel Roca-Martinez publications. 32,33 For each dataset and each prediction, every amino acid was further labeled to show whether it was a single (S) early folding or nonearly folding residue or whether it appeared at the beginning (B), middle (M) or end (E) of a given early folding or nonearly folding fragment (Figure S1). For each residue in each protein, a comparison was then performed for the overlap between the ef, bb and SIGb predictions in terms of true/false positives and true/false negatives.…”

A Perspective on the Prospective Use of AI in Protein Structure Prediction

“…One should note that disorder in proteins comes in different flavors, with conditional order that can arise from different experimental conditions. 107 High-confidence pLDDTs have been shown to sometimes correspond to residues that belong to disordered protein fragments in a monomeric unit that will fold conditionally, for example, when binding to another protein partner. 108,109 As a consequence, AF2 models should be taken with caution, as several studies show how they are likely to predict a proteinbound structure instead of its unbound structure in solution.…”

“…In that case, one can combine AF2 predictions with an additional tool based on the residues’ physicochemical properties, such as their hydrophobicity in the case of the hydrophobic cluster analysis (HCA), to unveil ordered segments that remain hidden from AF2. One should note that disorder in proteins comes in different flavors, with conditional order that can arise from different experimental conditions . High-confidence pLDDTs have been shown to sometimes correspond to residues that belong to disordered protein fragments in a monomeric unit that will fold conditionally, for example, when binding to another protein partner. , As a consequence, AF2 models should be taken with caution, as several studies show how they are likely to predict a protein-bound structure instead of its unbound structure in solution. − …”

Protein structuromics: New observations for translational medicine research in lung cance

“…We followed Grau l and Joel Roca-Martinez publications 43,44 . For each dataset and each prediction, every amino acid was further labelled whether it was a single (S) early folding or nonearly folding residue or whether it appeared at the beginning (B), middle (M) or end (E) of a given early folding or nonearly folding fragment (Supplementary Fig.…”