Graph representation learning for structural proteomics

Abstract

The field of structural proteomics, which is focused on studying the structure–function relationship of proteins and protein complexes, is experiencing rapid growth. Since the early 2000s, structural databases such as the Protein Data Bank are storing increasing amounts of protein structural data, in addition to modeled structures becoming increasingly available. This, combined with the recent advances in graph-based machine-learning models, enables the use of protein structural data in predictive models, with the goal of creating tools that will advance our understanding of protein function. Similar to using graph learning tools to molecular graphs, which currently undergo rapid development, there is also an increasing trend in using graph learning approaches on protein structures. In this short review paper, we survey studies that use graph learning techniques on proteins, and examine their successes and shortcomings, while also discussing future directions.

Drug repositioning in drug discovery of T2DM and repositioning potential of antidiabetic agents

“…The main sequence-based [48] input representations are international chemical identifier (InChI) and Simplified Molecular Input Line Entry System (SMILES), which use specialized syntax to encode three-dimensional chemical structures as strings of text [49] , [50] . Graph-based representations [51] , [52] can represent the spatial and structural relationships of proteins and biomolecules as well as the interconnected nature between different biological data [53] . Representation learning is mainly divided into supervised and unsupervised learning approaches to extract features of the input data for downstream training tasks.…”

Protein Representations: Encoding Biological Information for Machine Learning in Biocatalysis

“…An alternative approach to grid representations is to collapse the 3D protein structure to a graph representation where the structural information of the protein is encoded as elements and connections, designated as "vertices"/"nodes" and "edges", respectively (Fasoulis et al, 2021). Different detail levels can be employed when creating protein graphs, e.g., for atomistic resolution, features of each node consist of atom type and charge, while the edges represent the molecular bonds (Fasoulis et al, 2021). A more coarse-grained approach is the residue-level description where the nodes represent entire amino acids and the edges specify both the covalent and non-covalent interactions between the residues.…”

“…A more coarse-grained approach is the residue-level description where the nodes represent entire amino acids and the edges specify both the covalent and non-covalent interactions between the residues. For residue-level protein graphs, the node features can include physicochemical properties such as polarity and hydrophobicity (Fasoulis et al, 2021), or more advanced residue encodings such as evolutionary information or secondary structure (M. Li et al, 2023). Importantly, a graph is a non-linear data structure.…”

Machine Learning-Guided Protein Engineering

“…Graph-based protein representations have recently attracted considerable attention . A graph consists of a set of nodes linked by a set of edges.…”

“…Graph-based protein representations have recently attracted considerable attention. 57 A graph consists of a set of nodes linked by a set of edges. The nodes typically represent residues, atoms, or groups of spatially close atoms, while the edges usually correspond to chemical bonds, spatial proximity (contacts) between the nodes, or both.…”